Arsenic and Old Bookcloth: Identification and Safer Use of Emerald Green Victorian-Era Cloth Case Bindings

ABSTRACT Analysis of nineteenth-century, cloth-case publishers’ bindings at Winterthur Library revealed starch-coated bookcloth colored with the toxic pigment “emerald green” (copper acetoarsenite). While this pigment was widely used in Victorian home goods and apparel, its use specifically in bookcloth has not been formally explored. A survey of bookcloth pigments was conducted to identify which toxic elements may be present in bookcloth and to determine how many bindings were colored with emerald green. English-language books at Winterthur published between 1837 and 1900 were analyzed with x-ray fluorescence spectroscopy. When arsenic and copper were found together, Raman spectroscopy was used to confirm emerald green. The dataset was further expanded using The Library Company of Philadelphia’s significant holdings of cloth-case publishers’ bindings. Copper and arsenic were detected in 38 rare and circulating books in total. Based on this data, trends in emerald green bookcloth use include: consistently vivid green hue; stamped decoration; English and American imprints from 1840s to 1860s. Quantitative analysis revealed significant levels of arsenic in friable bookcloth colorant. Emerald green books in the Winterthur collection were rehoused in zip-top polyethylene bags with hazard labels and circulating books were moved into the access-controlled rare book collection.


Introduction
The Winterthur Poison Book Project seeks to identify potentially toxic pigments used to color nineteenth-century bookcloth and to provide recommendations for mitigating the risks associated with the handling and care of such cloth-bound books. This article focuses on one toxic pigment in particular: emerald green (copper acetoarsenite).
In early 2019, during exhibit-related examination and treatment of the volume Rustic Adornments for Homes of Taste (Hibberd 1857), the suspiciously bright green hue and friability of the colorant prompted pigment analysis in Winterthur's Scientific Research and Analysis Lab. The bookcloth colorant on Rustic Adornments was confirmed to be copper acetoarsenite, or emerald green. The presence of a friable toxic pigment on the exterior of a book in Winterthur Library's collection caused concern for the safety of library staff and patrons alike, the more so because many Victorianera, cloth-case bindings are housed in Winterthur's circulating collection. This concern prompted an analytical survey of the collections, in order to understand better how many of the library's case bindings might be covered in potentially toxic bookcloth, and to determine the required scope of safe handling and storage precautions. Further analysis was conducted at The Library Company of Philadelphia in order to expand the initial data set. This data and attendant contextual research informed Winterthur's development of policies and procedures for the safer handling, conservation treatment, and storage of emerald green bookbindings.

Bookcloth: history and production
Commercially successful bookcloths were a closely guarded trade secret during the nineteenth century, and our current understanding of their materiality and manufacture is still incomplete. The occasional use of dressmaking fabrics for covering books occurred prior to the invention of commercial bookcloth in the 1820s. However, these fabrics were unsatisfactory as a bookbinding material. They allowed adhesive to ooze through and stain the outer surface of the cloth and lacked durability as book coverings, fading, fraying, and splitting catastrophically with use (Miller 2014). The Poison Book Project builds on work by Tomlinson and Masters (1996) elucidating late nineteenth-century bookcloth manufacturing techniques. Their research is based on the operations of the Winterbottom Book Cloth Company, located in Greater Manchester, England, UK, a bookcloth manufacturer that dominated the industry in the last quarter of the nineteenth century. At Winterbottom, bookcloths were colored by applying a surface treatment rather than dyeing cloth fibers directly. Bleached cotton cloth was padded, meaning a colored starch slurry was scraped into the interstices of the weave, imparting color while also making the cloth more impervious to adhesive squeeze-through during the bookbinding process. Bookcloths were then backfilled with a material composed of colorant mixed with starch and other fillers. The backfill material sits on the surface of the cloth, and serves the function of both colorant and coating. This technique results in brilliantly hued bookcloths, but their surface can be vulnerable to mechanical abrasion. While Tomlinson reveals the manufacturing process used at Winterbottom, his research does not indicate the specific pigments or dyes used as colorants. Recipes for various green shades of Winterbottom cloth manufactured in 1903 reference adding a quantity of "mid green" or "medium green" to a starch and gum slurry without specifying the added compound (Tomlinson and Masters 1996). This vagueness in the documented recipes may be one means by which book cloth producers kept details of the process hidden from their employees, out of fear that an experienced workman might leave to set up a rival manufactory.
Beyond Tomlinson's work, most studies of bookcloth have focused on describing embossed cloth grain patterns for identification and descriptive purposes. Since the 1930s, contributions to this effort have been made by a number of book historians, culminating in Krupp's comprehensive inventory of early nineteenth-century bookcloth patterns (Sadleir 1930;Carter 1932;Tanselle 1970;Spawn and Kinsella 2002;Krupp 2008). Most recently, Lundblad's study of the material culture of cloth publishers' bindings in Sweden briefly considers the innovation of bookcloth within a larger social, economic, historical, and design context, although she relies entirely on Tomlinson for a technical understanding of the material itself (Lundblad 2015).
Winterthur's Rustic Adornments for Homes of Taste displays a binding ticket for Westley and Co., a prolific London bindery in the mid-nineteenth century. According to a contemporary article in Penny Magazine (1845), "A Description of Westleys & Clark's Bookbinding Establishment" (the predecessor to Westley and Co.), bookcloth was purchased in rolls and stored in a cloth warehouse, where it was cut down into pieces of several yards each before being handed over to the cloth case makers. Gas-jet powered embossing machines impressed patterns on the precut pieces of bookcloth to hide the weave. The article also notes that gilding was reserved for a "higher class of bound book," which speaks to the consumer demographic for many of the emerald green bindings identified during this study (Penny Magazine 1845). Neither contemporary reports on bookbindery processes nor more modern scholarship elucidates the colorants used in nineteenthcentury bookcloth manufacture.

Arsenic-rich green pigments
The nineteenth-century craze for brilliantly-hued arsenic-based greens in material goods is well documented (Whorton 2010). These pigments could be found in everything from apparelsuch as gowns, hats, and shoesto children's toys and wallpaper. The passion for emerald green required some degree of cognitive dissonance on the part of Victorian consumers, who also purchased and used the arsenic-based green pigment as a rat poison and agricultural insecticide. Emerald green is a bright, brilliant green first synthesized around 1800, and commercially produced in Schweinfurt, Bavaria beginning in 1814 (CAMEO 2020). 1 Its toxicity was realized soon after, but it continued to be sold and used widely. Emerald green is more lightfast than the yellowish-hued Scheele's green (copper arsenite), but is still susceptible to sulfurous air pollution, which causes the compound to oxidize and darken. Its popularity as a green colorant in everyday consumer products persisted in Victorian England and the United States throughout the nineteenth century in spite of a lively, widespread debate about its health risks in English and American newspapers from the early 1860s onward. Alarmed medical practitioners and consumer advocacy groups warned women, in particular, of the danger lurking in their evening gowns, while economists and industrialists countered that a ban on the popularand lucrativecolor might cause irreparable harm to the British economy (Whorton 2010).
In 1862, a New York publication featured a warning about the hazard of emerald green ballgowns under the sensational title "The Dance of Death" (Frank Leslie's Illustrated Newspaper, May 17, 1862). The article claimed that "the color is loosely laid on with starch, and comes off by the slightest friction in clouds of dust … One physician satisfied himself that from a lady's dress of this kind not less than 60 grains, sufficient to kill 30 persons, powdered off in a single evening during the lady's dance at a ball." While this claim cannot be substantiated, it does speak to the public climate in which such goods were manufactured and consumed.
The description of the cloth from which this gown was made also resonates with a technical understanding of the manufacture of bookcloth, during which colorant is applied in layers of starch-containing binder, resulting in a potentially friable surface.
While most of the public notoriety focused on ballgowns, millinery, and home decor, emerald green bookbindings have been mentioned in various contemporary reports and articles, although mostly as an unelaborated afterthought. On September 5, 1863, The Sheffield and Rotherham Independent reported on investigations undertaken by the Ladies' Sanitary Association and the medical department of the Privy Council into the "injurious" nature of arsenical greens as used in artificial flowers, confectionary, ball gowns, toys, paper hangings paper wrappings, and "as covers for tickets, boxes, or books." A more comprehensive exception is a specific case of poisoning caused by the glazed green paper covering a children's book: Case of poisoning by an arsenical book cover -A child in Troy, N.Y., while playing with some water-color paints, used a small book, attracted by the bright green color of its cover, for a palette. He mixed the paints on the cover of the book for some time. Then he was suddenly taken with convulsions. Physicians, who were hastily summoned, declared that he had been poisoned. They administrated antidotes, but the child went into convulsion after convulsion, and it was only after three days incessant labour that the physicians saved his life. They afterward investigated the manner of his being poisoned. They discovered that the dye with which the bright-colored book was covered contained the poison. In wetting the paints on the book-cover, the child had innocently wet also the dye, and soon transferred some of the poison to its own lips. (Carr 1883, 47)  Wood notes that common arsenical domestic products included "Paper … for covering children's and other books; … and book-cloth and fancy bindings," the latter term aptly describing the highly decorated covers (Figure 1) of the majority of emerald green cloth bindings identified by Winterthur conservation staff (Wood 1884). Further investigation into glazed bookbinding papers from this period is well warranted, but the scope of the current study centers around the dearth of information on the use of arsenical green pigment in bookcloth.
Arsenic is a not only a constituent of copper arsenic green pigments, but is also found in the yellow pigment orpiment (As 2 S 3 ). The first published study of arsenical bookbindings describes arsenic-rich paint used in sixteenth-and seventeenth-century European tawed skin bindings; analysis suggested the presence of orpiment (As 2 S 3 ) mixed with indigo and quartz to produce a green color (Delbey et al. 2019). Orpiment was a popular yellow pigment in the European painting palette throughout the sixteenth to eighteenth centuries (CAMEO 2021b). While the primary focus of the Delbey article is the technical analysis and identification of the green colorant, important suggestions are also made for the safe handling of arsenical bookbindings in the research library environment.

Analytical survey
The primary purpose of a survey of bookcloth in the Winterthur Library was to determine whether Victorian-era bookcloths colored with toxic pigments were a rarity or a fairly common hazard that persists in collections today. The analytical survey was limited to English-language, cloth-case bindings published during the Victorian era . This scope of language and dates was determined for a number of reasons. England dominated commercial bookcloth production in the nineteenth century, with the material widely used in the British Isles and North America; limiting the survey to English-language publications made identification of appropriate cloth bindings in the collection a more manageable task. The dates of the Victorian era neatly span two key moments in the history of bookcloth. First, while commercially viable bookcloth was invented in the 1820s, the industry did not develop into mass production until the mid-1830s, which coincides with the beginning of the Victorian era. Second, by 1900, brightly colored, lightfast, coal tar dyes had superseded many of the textile colorants popular in the nineteenth century, so potentially toxic pigments are more likely to be found in cloth bindings before this date.
Using portable X-ray fluorescence spectroscopy (pXRF) for elemental identification 2 (Figure 2), interns at Winterthur initially analyzed 200 books in a range of bright colors, including reds, yellows, greens, and blues. The survey focused on the presence of heavy metals, which are widely considered potentially toxic across many disciplines concerned with public health; the heavy metals identified in bookcloth include arsenic, chromium, copper, iron, lead, and mercury. Where arsenic and copper were present, the elemental analysis was followed with Raman spectroscopy to confirm the molecular structure of the compound 2 ( Figure 2). The bookcloth covering the front board of each analyzed book was photographed for reference. This initial survey revealed copper acetoarsenite (emerald green) to be the pigment of most immediate concern, so after the first 200 books, analysis continued on books bound in green bookcloth only to work through the collection more efficiently. At Winterthur Library, a total of 406 volumes, including all green case bindings in both the circulating and rare book collections, have been analyzed with pXRF. Approximately 80 green case bindings in the American and British publishers' binding collections at the Library Company of Philadelphia were also analyzed with pXRF. The Library Company of Philadelphia shelves its Americana collection chronologically according to imprint date, so volumes were selected for analysis by a visual scan of the shelves in the appropriate date range.
In total, nearly 500 volumes have been analyzed across the two institutions, approximately 350 of which are covered in green cloth. Of those, arsenic and copper were detected in 38 volumes, which constitutes over 10% of the green cloth-case bindings analyzed. Emerald green cloth-case bindings range in size from large quartos to petite duodecimos (Figure 3). The current data set is still too small to see any trend in publishers, but there are other helpful commonalities. The majority of emerald green bindings have publication dates in the 1850s (Figure 4). A few were published in the 1840s, but this does not necessarily indicate that emerald green bookcloth was being manufactured and used in the 1840s, because remaindered books may have been bound later. Likewise, already-purchased stores of bookcloth may have continued to be used by binderies after bookcloth manufacturers ceased using emerald green as a bookcloth colorant. A time-effective strategy for those considering searching for emerald green books in their own collections might therefore start in the 1850s and work out in either chronological direction. The majority of the emerald green bindings identified in this survey are highly decorated, with gold titling, gold and blind blocking, and often gilt edges. The bookcloth over the boards remains vividly green, with no sign of insect damage. The condition of emerald green spine cloth shows more variation in color, likely from exposure to oxidizing air pollution as the book sits on the shelf. The survey also revealed trends in green bookcloth use overall. In emerald green bookcloth, copper acetoarsenite was the only colorant present. Non-arsenical green bookcloths analyzed with Raman spectroscopy were found to contain Prussian blue and chrome yellow pigments, a mixture more commonly known as chrome green. (Figure 5).

Alternative method of green pigment identification
Seven samples removed from bookcloth that had been previously analyzed by XRF and Raman spectroscopy were analyzed by polarized light microscopy to explore the potential of the technique for the reliable identification of emerald green. The books were selected and sampled by researchers other than the microscopist in order to emulate the analysis of unknown pigments. Examination in plane-and cross-polarized light conclusively and correctly identified emerald green as the colorant in three of the samples, and found a mixture of Prussian blue and chrome yellow in the other four samples. While all optical characteristics observed in the samples were considered in the interpretation of the results, the presence of spherulites was the most distinctive diagnostic feature in the identification of emerald green. Tips and recommendations for sampling, sample preparation, and analysis by PLM are available (Supplementary Material 1). L84t. Small amounts of iron are also detected in XRF, which could be due to surface grime. A reference Raman spectrum of emerald green is shown (B).

Quantitative analysis
The qualitative survey results revealed that emerald green was not uncommon in mid-nineteenth-century bookcloth, but quantitative analysis was needed to grasp more fully the risks associated with emerald green bookcloth. A destructive sample, in triplicate, of 1 cm 2 bookcloth taken from underneath the pastedown of the octavo-sized book Rustic Adornments for Homes of Taste (Hibberd 1857) was sent to the University of Delaware Soil Testing Lab for quantitative elemental analysis using inductively coupled plasma-optical emission spectrometry. The Soil Testing Lab also analyzed "pick-up tests" of dry 2 ′′ × 2 ′′ cotton pads swiped once along the length of the book's front board and dry cotton swabs rolled lightly along the length of the book's front board intended to simulate common handling. Triplicate samples were weighed to ensure they weighed above the minimum 1 g required for analysis. Samples were placed in 10 mL of concentrated nitric acid. Samples were then digested by EPA Method 3051A using a CEM MARS5 microwave digestion system. The digest was then diluted to 50 mL using deionized water. Digests were analyzed for arsenic using a Thermo 7600 ICAP 7600 Duo ICP-OES. Reference soils from "High-Purity Standards" were digested and analyzed with every sample run as they had the analytes in question in measurable quantities. These references served to ensure that the sample had digested properly and to check the recovery of the elements being measured. These reference samples were chosen because they were at least as resistant (and possibly more so) as the bookcloth, and good recoveries were achievable for both of them. The quantitative analysis indicated that the bookcloth contains several times the lethal toxic dose of arsenic for an average-sized adult, as little as 2 mg/kg of body weight (Table 1) (Gehle 2010). The pick-up tests resulted in a significant, measurable amount of arsenic offset from the dry bookcloth. Standard deviations for pickup tests are high, showing the poor reproducibility of the sampling method and the venting of the closed microwave digestion vessel during one sample run. Venting can result in slightly lower values if the steam that vents carries any of the constituents being measured with it. A third pick-up test using nitrile gloves was also conducted, but the Soil Testing Lab protocol was unable to digest the nitrile, and therefore provided no results.
There are limitations to the interpretation of these results. This singular binding may not be representative of all emerald green bookcloth. Not all emerald green bookcloth may be this friable. It is also impossible to determine how much pigment this binding may already have shed over its nearly 200-year history. During the sampling for PLM analysis described in 2.1, the bookcloth colorant on the three emerald green bindings, all printed by different English publishers, was observed to be generally more friable than the chrome green colorant, which appeared to be more effectively bound to the bookcloth. Whether this trait is a result of manufacturing technique or the nature of the pigment itself is unknown. Further research is needed to explore how the chemical deterioration mechanisms of copper acetoarsenite may impact binding media, and how external factors such as environmental storage conditions may influence the colorant's friability in bookcloth.

Health and safety discussion
Conversations about this research with experts from various fields associated with health and safety have provided a broader context for interpreting the degree of risk present as outlined by the Agency for Toxic Substances and Disease Registry (Gehle 2010). Forensic toxicology, epidemiology, and industrial hygiene are related disciplines which consider risk from differing perspectives. According to forensic toxicologist From the perspective of pathology, he considers arsenical bindings to be low risk; however, he also acknowledged that institutions may need to act with an abundance of caution because of issues surrounding legal liability. Epidemiologist Dr. David Goldsmith Department of Environmental and Occupational Health, George Washington University (personal communication, October 30, 2019) focused less on fatality as the primary risk factor, drawing attention instead to the potential for materials people interact with in their daily lives to trigger serious, long-term health burdens. Goldsmith suggested research into arsenical bookcloth should be communicated not only to conservators and librarians, but also to epidemiologists, because potential problems from long-term, low-grade exposure to arsenic could be a public health concern. Safe handling and storage protocols for arsenical bindings at Winterthur Library are based on consultation with industrial hygienist and University of Delaware Director of Environmental Health and Safety Michael Gladle (telephone conversation, December 13, 2019). From Gladle's perspective, there are no safe exposure limits for copper acetoarsenite, so limiting direct contact, inhalation, and ingestion of bookcloth pigment insofar as possible is essential.
Emerald green bindings present a risk to library staff and users that should be taken seriously. While every institution must work out its own logistics for meeting the goals of safe storage, handling, and treatment, the approaches being explored at Winterthur Library may provide a useful model. First, steps have been taken to restrict circulation of arsenical bindings. Arsenical books previously housed in the circulating collection at Winterthur have been moved into the rare book collection. Whenever possible, researchers will be encouraged to use a digitized surrogate instead of the original. Researchers with a compelling reason to handle original emerald green bindings will do so in the controlled environment of the Winterthur Library rare book reading room, under the supervision of trained library staff. All library staff have been trained in safe handling practices, and a written policy is posted for easy reference in staff areas of the library, including in the rare book stacks at the location where the arsenical bindings are stored.
Other precautions in place at Winterthur recommend individuals wear nitrile gloves when handling arsenical bindings; take care to avoid ingestion or inhalation of pigment dust; and avoid touching the face, eating, or smoking until hands can be thoroughly washed, even if gloves have been worn to handle bindings. Arsenical bindings will be used only on hard surfaces which can be wiped down with a damp, disposable cloth. A nonpermeable "placemat" of polyethylene or polyester will be draped over book wedges, cushions, or futons during use. Additional guidelines for researcher use of arsenical bindings and the potential use of waivers are still under discussion among conservation, library staff, legal counsel, and other stakeholders at Winterthur.
The protocol developed at Winterthur for the conservation treatment of arsenical cloth bindings adheres to recommendations from University of Delaware Environmental Health and Safety. In addition to wearing nitrile gloves and observing the conscientious hygiene practices already described, conservators should carefully consider the environment in which they plan to treat books bound in arsenical bookcloth. Moisture introduced by adhesives could cause the arsenic to migrate and offset in greater concentration. Exposure to a localized heat source or contact with acids can also cause copper acetoarsenite to emit highly toxic arsine gas (NOAA n.d.). Therefore, best practice indicates working under a certified chemical fume hood. A second choice would be to work in a ductless particulate  hood with a combination HEPA/charcoal filter. A respirator with organic solvent and particulate filters should be considered only as a last resort. Conservation and library staff at Winterthur agreed that enclosures for arsenical bindings should be used to prevent casual touching and contain any pigment shedding. First, standard enclosures already in use were considered: the CoLibri book jacket system, and corrugated clamshell boxes. CoLibri jackets are made of polyethylene, an effective choice of material as a barrier for emerald green pigment. However, CoLibri jackets have a large gap that exposes the head and tail of the spine as well as a portion of the board edges, where colorant is most vulnerable. Paperboard boxes were also evaluated and rejected, as they could collect and then release shed pigment unpredictably during handling.
Ultimately, staff consensus settled on the humble, but perfectly serviceable, polyethylene zip-top bag. The manual logistics of sliding materials in and out of the bag encourages extra attention and care. The transparency of the polyethylene allows easy monitoring of pigment shedding, and the bags are easy to label. Bagged books can continue to be shelved upright, and have a tight zip-top seal that keeps pigment in and water out in the case of a water disaster. The zip-top bag differs enough from other standard enclosures at Winterthur Library to signal that something is notable about these contents; however, if desired, a bagged book could additionally be boxed.
Recommendations from UD Environmental Health and Safety suggest storing arsenical books together in one area with clear shelf signage, in addition to individual labels on each enclosed item. Winterthur Library already has a tradition of shelving certain collections items together based on physical attributes. For example, folia and miniatures collections are housed in designated shelving areas rather than being interspersed throughout the rest of the collection. A letter code at the end of the call number alerts library staff and users to the item's special location. Re-housing arsenical books together in one storage area easily fits into this pre-existing practice. A clearly labeled shelf in the rare book vault has been designated for Winterthur's 10 arsenical case bindings, which further minimizes risk to salvage staff in case of a collection emergency. It should also be noted that a mold outbreak presents additional risk when emerald green bindings are affected, because certain fungi can digest copper acetoarsenite, releasing highly toxic arsine gas (Gosio 1897;Challenger 1951). Staff responding in such a situation should wear appropriate PPE, including a full-face respirator (NIOSH 2019). Guidelines for safe handling of arsenical materials should also be added to the institutional disaster response plan.

Ongoing and future research
A primary goal of this project is to alert libraries and private book collectors to the potential hazard of arsenical cloth bindings in their collections. Without access to analytical instrumentation to identify arsenic, reliably detecting emerald green bindings presents a challenge.
Arsenic spot testing is commonly used as a means for identifying natural history specimens treated with arsenic trioxide, a white amorphous powder commonly used as a preservative and pesticide (Knapp 2000;CAMEO 2021a). It is likely that such tests could be used successfully to identify the presence of arsenic in bookcloth. However, microchemical spot tests should be performed only in an appropriate laboratory setting by individuals trained in the safe handling and disposal of the required chemical reagents, which are themselves hazardous (Marte, Pequignot, and von Endt 2006). Given the increased risk of exposure during the sampling of potentially arsenical compounds, as well as general laboratory safety protocols, spot testing is not recommended for library staff or private collectors who do not have the appropriate facilities or training.
An emerald green color swatch bookmark has been developed at Winterthur as a public outreach tool, with the hope that it may also serve as a tool for the non-instrumental, visual identification of books which might be bound in emerald green bookcloth. While the bookmark has been shared freely with the public upon request, instrumental data collection to support or contradict its effectiveness as a subjective tool for visual identification of emerald green bookcloth is still underway. The development and reliability of the bookmark as an identification tool will be reported upon in a future publication.
In order to share interpreted trends in the use of emerald green bookcloth and safety tips related to arsenic more broadly, Poison Book Project information is maintained on the Winterthur Wiki. Although a wiki page is gray literature, this web-accessible format allows nimble updates to public-facing information about the project as it evolves, including instructions for requesting a bookmark and for contributing data to the project. 3

Poison book project database
The Winterthur Wiki provides an access point for the Poison Book Project online database. The database provides bibliographic information (title, author, imprint, publication date) for books bound in emerald green bookcloth. When permission has been granted by institutional or individual owners, the collection is also identified. The means of identification of the emerald green bookcloth is also provided, in acknowledgment of the varying degree of confidence of different methods. Contributors outside of the Winterthur research team who wish to contribute to the database are asked to submit a form containing the required data, along with corroborating evidence to support the identification. Corroborating evidence may be a photomicrograph using the PLM protocol discussed in Section 2.1, or spectra from XRF or Raman spectroscopy. This information is reviewed by project researchers before being approved and added to the database.

Conclusions
Ongoing analysis, archival research, and outreach to libraries, private book collectors, and book dealers continues for this project. Next steps include the expansion of the arsenical bindings data set through the contributions of other institutions. While collaborations and archival research have been delayed by the COVID-19 pandemic, planning for post-COVID, on-site analysis and research continues. The project also continues to collect data in the Winterthur collection about other pigments of concern, including chromium-based, mercury-based, and lead-based pigments.
While this initial study focuses on the specific risks associated with arsenic-based emerald green pigment, generalized lessons can also be drawn from this research. First, allied disciplines related to health and safety have differing perspectives and risk tolerances, so consulting experts from multiple adjacent fields provides breadth of context for decision-making. Second, a cookie-cutter approach to the logistics of storing, handling, and serving potentially toxic materials to library patrons will not work for every institution; involving stakeholders in a conversation about desired outcomes based on core safety guidelines can achieve a tailored solution that works. Finally, libraries need more materials research to identify hidden hazards in their collections. Research partnerships and interinstitutional knowledge-sharing is vital for protecting library staff and users as well as collection materials themselves.

Notes
1. There are many common names for arsenic-based green pigments, and these were often used interchangeably (and therefore inaccurately) by the contemporary public. Some historical sources also mistakenly conflate Scheele's green (AsCuHO 3 , copper (II) arsenite) with the various other names for emerald green (Cu(C 2 H 3 O 2 ) 2 ·3Cu(AsO 2 ) 2 , copper (II) acetoarsenite), such as Schweinfurt green, Paris green, Vienna green, Mitis green and King's green; however, Scheele's green and emerald green are distinct compounds. Paris green was a commercial name commonly used when emerald green was sold as a pesticide. Emerald green was the name used to describe the compound copper acetoarsenite by the German chemists who first synthesized it, and these two terms will be used interchangeably here (Fitz-Hugh 2019). 2. Elemental analysis was performed with a handheld Bruker Tracer III-SD XRF spectrometer using a rhodium tube (40 kV high voltage, 9.6μA anode current, 25μm Ti/305μm Al) for 30 s live time irradiation. A zero-background plate (design courtesy of Institute for the Preservation of Cultural Heritage, Yale University) was placed behind the cover to mask elements within the textblock. Spectra were interpreted using the PXRF1 software. When arsenic and copper were found together on Winterthur volumes with pXRF, Raman spectroscopy was used to confirm copper acetoarsenite ( Figure 2