The Pregnant Image: Ethical Considerations in Archival Practices”

Abstract

This project, The Pregnant Image, documents the collaboration between Emily Zarse, an artist-curator, and me as we consider the archival ethics surrounding a collection of printed photographs. The photographs, which depict pregnancy, labor, and postnatal care, were individually salvaged by Zarse from eBay and Etsy. These already vulnerable photos were purchased from sellers that used harmful tags in their descriptions of the subjects, oftentimes fetishizing the contents of the photographs. After she had acquired over six hundred objects, Zarse was unsure how to preserve these photographs without perpetuating the harm already done to them. Over the course of the semester, I built a unique metadata profile to describe the first sixty-seven objects in the collection. This digital project was created using CollectionBuilder, an open-source framework for creating digital collections that are metadata driven and powered by modern static web technology. I customized CollectionBuilder’s features so the site could host an abstracted view of the archive without reproducing these vulnerable images. Zarse and I decided to upload scanned reproductions of photographs that naturally obstructed the facial features of the subject(s), but we currently choose to keep images with identifiable subjects private. Visitors can view the first ten envelopes of the collection through the “Browse” tab. Our conversations are ongoing; the “About” drop-down menu on the site details a timeline of our collaborations, an essay about the collection, and technical information about the digital archive.

Introduction and Origins of the Data

Women’s stories suffer from what folklorist Brian Sutton-Smith calls the “triviality barrier”, especially when it comes to pregnancy and birth. Emily Zarse is a fibers artist and curator located in Bloomington, Indiana whose work centers motherhood and pregnancy. After receiving a grant through the Bloomington Arts Commission, she began searching for authentic photographs depicting pregnancy as personal research for her next project. Carmen Winant’s 2018 installation in the Museum of Modern Art (MoMA) titled My Birth had catalyzed an internal conversation in Zarse. The work, installed on two parallel walls of a hallway, is composed of over two thousand images of women in labor and childbirth. Winant sourced these photos from books, magazines, the inventory of used bookstores, and garage and estate sales. At first, Zarse was moved by the piece’s ability to amplify motherhood and offer a diverse and visible depiction of pregnancy through photography. But as she sat with it longer, she began to ask, ‘Why are none of these photographs labeled? Where is the consent to the exhibition of these photographs?’. Together, Zarse and I are applying this critical review of a niche collection of photographs, which began with Winant’s work, onto the archive that Zarse has now curated herself.

Over the course of the spring 2023 semester, Zarse purchased photographs she found on eBay and Etsy that depict pregnancy. She remarked that finding these photographs reinforced the trivialization of the topic after realizing that she quickly exhausted search words that would successfully produce photos of her interest. “Pregnant woman” served as the most successful search term, followed by “pregnant” and “baby”. She also used the terms “belly”, “pregnant belly”, “breastfeeding”, “hospital birth”, “home birth”, “placenta”, “midwife”, and “OBGYN”. Throughout her search, she noticed that many of the sellers of the photographs used harmful tags to describe the images, some of which included fetishized and objectifying diction. Based on the size of some of the sellers’ collections and the variance of the subjects depicted within, it is unlikely that the original owners are also the sellers of these photos. Over the summer, Zarse began questioning whether these photos should be preserved and archived, and if so, what the best practices would be for the preservation of this vulnerable collection. From this phenomenon, there emerge key research questions: How does one ethically archive a collection of vulnerable photographs? How do we give dignity, agency, respect, and restoration to the historically vulnerable and marginalized subjects in the archive? How can digital technologies be used to harm and heal sensitive materials? As a colleague and admirer of Zarse’s work and scholarship, I offered to build and analyze a dataset of these photos. In this project, my own interest in life cycle rituals, particularly birth, marries my interests in ethical archiving and creating static-site driven digital community archives that employ non-custodial archiving practices and minimal computing methodologies.

Foundations of the Collection

Like Winant’s collection of photographs, the materials in Zarse’s archive span several decades. Given that the sellers are in different geographic pockets of the Unites States, we assume that the photographs mimic a similar dispersion. Some sellers listed the items as “found photos”; we suspect that the sellers have either purchased these photos through estate sales, garage sales, online, or were “found” by more ambiguous means. Over the course of her research, Zarse found that she became more partial to purchasing photographs when she saw harmful language being used in their corresponding descriptions. As a result, the collection is comprised of many of these photographs. As an artist, Zarse is particularly interested in poses commonly employed by photographed pregnant women. For example, the quintessential side profile pose captures the expectant mother as she is accentuating the proportions of her pregnant belly by positioning herself perpendicular to the camera lens. Zarse is also curious about the placements of hands in photographs of pregnancy, which are commonly found on top of or below the belly. The role of community support is additionally of interest; Zarse’s research will analyze when pregnant people are photographed in solitude or communion. In order to create a dataset that would aid Zarse’s research, the contextual background of the collection, the presence of additional subjects in the photographs, and their implied roles needed to be represented in the final product.

The librarian’s role is to amalgamate the patron’s requests and produce search results that aid them in their queries. In order to transform Zarse’s research interests into identifiable data, the dataset needed customized fields to identify and describe Zarse’s research interests: like any handwritten script found on the back of the photographs, the subject’s stage of pregnancy, the seller’s description, the seller’s racial description of the subject, a description of the subject’s pose, whether the subject’s face is obstructed, the presence of support in the photograph, and the support’s implied role. This dataset also serves as an intentional abstraction from the original photograph. In the digital archive that hosts this dataset, one can read the profile for every photograph in the collection and garner a sense of its subjects, setting, and vulnerability without having to look at the original film. This method of digital archiving aims to eliminate the anxieties of resuming the film’s circulation on the internet.

Data Ethnography and Determining What Counts

There are theories abound when it comes to best practices of understanding and manipulating data. Johanna Drucker’s idea that, “Data are capta, taken not given, constructed as interpretation of the phenomenal world, not inherent in it”, recognizes that interpreting datasets is an act of phenomenological bias. Steve Braun explains Drucker’s sentiment that, “whereas data presumes observer independence and absolutism, capta acknowledges the situatedness and observer codependence of the interpretive act.” The steward of the data, the data itself, and the data visualization are all in conversation with one another, meaning that none of these actors is fully individualized. At the same time, the data is at the mercy of its stewardship. The interpretation of data lies in the conscious or unconscious biases of its interpreter. Digital humanists can unpack the intricacies of the data’s captivity through the analytic tool of data ethnography. In “Attending to the Cultures of Data Science Work”, Lindsay Poirier encourages opportunities to extend and implement reflexive protocols beyond dataset creation and specifically calls on data science programs to implement ethnography, hermeneutics and critical analysis in their research development.

Data ethnography combines principles of the humanities, particularly the research methodology of ethnography, the act of immersing oneself in the life history of the individual interest, with the big-picture scope of analysis found in data science. It “is a means of understanding social worlds within data consumption…involves how users consume data, how they are circulated and how data shapes how people interact and live their lives” (Knox, 2018). Data ethnography can involve any of the methods associated with ethnography, including analyzing the contextual information surrounding the data, interviewing and observing who work with data, analyzing the policies and procedures surrounding data, participating in data practices, and conducting “close readings” of data, specifically in regard to how terms are defined and categorized, what gets counted in the dataset, and who decides what gets counted. By performing data ethnography, researchers can gain a richer understanding of the ways humans encounter data, and this analysis can be extended to actors at all levels of the data collection and development process: users, producers, laborers, funders, etc. Insights from data ethnography help researchers pinpoint areas of concern with a particular dataset. From these insights, ethnographers can suggest new strategies or approaches that can be implemented to solve future issues.

When considering the contextual information surrounding the dataset of analysis, researchers must keep in mind that data are not neutral or objective, but rather products of social relations. This context is essential for conducting accurate and ethical analyses (D’Ignazio & Klein, 2020). When these principles are applied to Zarse’s collection of photographs, the predatory nature of the sellers becomes quite visible. Perhaps the phenomenon of Etsy and eBay photo sellers seems innocuous due to the sheer amount “found photos” that sellers list on these free market sites, but there is something unsettling about the consumption of these photographs and their popularity driving sellers to invest in the search and capture of these “found” materials. To make a profit from these listings, sellers must anoint them with titles and tags that target consumers most likely to purchase the photographs, which is a subjective judgment that the seller places on their hypothetical customers. When Zarse searched these platforms for photographs depicting pregnancy, she was met with seemingly neutral listings, but also found items listed with language that sexualized the racial identity and pregnancy status of the subject. Zarse remarked that seeing these listings elicited a strong enough emotional reaction that she felt compelled to purchase them as a means of salvaging the photograph from a consumer that might actively searching for photographs with sexualized and fetishized content. After several collaborative meetings with Zarse, I began to re-home the collection materials into plastic photo archival sleeves and three-ring binders. As I transferred images from the envelopes, I would briefly study their contents. In the instances where sellers included a printed copy of the listing titles that Zarse purchased; I would attempt to match the image with the seller’s perspective of the ephemera. Hours into this process, I noticed tension in my stomach and a heightened sense of anxiety. After meeting with Zarse, she confirmed a similar effect when handling the materials in the collection, akin to a haunting, a “seething presence” that impacted, shaped, and meddled with our perceptions of the photographs (Gordon, 2008). It was not until I was visually and tactilely engaged with the contents that I pieced together the contextual bias and predatory scheme for profit that these sellers employ.

I provide this example to illuminate positionality practices and the dimensions of impact on equity research. Positionality captures dynamic individual presentations that are defined by socially constructed dimensions, but it can also reflect the choices researchers make and how they are informed by individual discourse-mediated identity and social interactions (Secules et. al, 2020). Identity is a combination of how one identifies and how one is perceived by society, and Secules et. al argue that research topics are inherently related to the experiences and positionalities of the researcher themself, creating an interplay of distance and proximity as one tackles analysis. As digital humanists, we have a responsibility to consider how our perspectives and experiences influence the ways we engage with research. Zarse herself is a mother and her works centers around motherhood and pregnancy because of her identity and positionality. It should be stated that, situated in her artistic practice, her emotional reaction to the exploitation of these images is correlated to her perception of pregnancy and vulnerability, which is phenomenologically tied to her own experience with the prenatal, labor, and postnatal stages of her life. As the digital archivist of these materials, my own positionality comes into play with the collection. Although I am not a mother, my interests in data feminism, cyborg theory, and ethical archiving are all at play in this project. My etic perspective as a White women with a negligible medical history of pregnancy is naturally in communication with the collection materials that depict the pregnant experience of expectant mothers both within and outside of my own socially constructed racial identity.

When considering the researcher’s influence on data collection and analysis, Secules et. al also discuss marginality, its relativity, and intersectional status. They maintain that a sense of privilege may inform a sense of duty to conduct research that dismantles a system that privileges the researcher. In the application of this photo archive, Zarse and I both view ourselves as researchers that are privy to the exploitation of the images. As researchers who are both White women, we are aware of the afforded and disadvantaged perspectives we conjure as it relates to this collection. Our positionality as women certainly draws us to the photos, and our knowledge that the most vulnerable subjects in the archive, that is, the only subjects whose race is explicitly defined in the titles and tags of the eBay and Etsy listings, are Black and Brown women. This will understandably call for an intersectional approach to the data which extends beyond a theoretical approach limited to a singular gendered, racial, or class analysis. We hope these considerations will be at the forefront of the user experience as they navigate through the materials within.

Race and Maternal Mortality

Zarse’s research also intersects with race, particularly as it pertains to the maternal mortality rate among Black women in America. A photograph is a permanently frozen moment in time, and in the case of pregnancy, the photograph captures a liminal state in the pregnant subject’s lifespan. In “Regarding a Pain of Our Own: Jazmine Headley, Portraiture, and the Sorrow of Black Motherhood” (a play on Susan Sontag’s “Regarding the Pain of Others”), Brie McLemore posits whether the medium of photography can seek to rectify harm that has been done to Black motherhood. The World Health Organization classifies maternal mortality as “the death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the duration and the site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management, but not from accidental or incidental causes”, which adds complications to the classification of a maternal death. Adding to further complication of a just representation of maternal death, the Center for Disease Control’s (CDC) data on maternal mortality is sourced from the National Vital Statistics System mortality file and only features three racial categories in its description, White, Black, and Hispanic (2024). In 2017, the CDC reported that Black women were over three times more likely than White women to die from pregnancy or child-birth related causes, but that neither education not income level impacted these fates.

The impact of these racial, educational, and class variables are painstakingly evident in Serena Williams’ birth complications. In 2017, Williams, one of the greatest tennis players of all time, almost died in childbirth due to complications from a pulmonary embolism, a sudden blockage of an artery in the lung, a condition she had previously experienced with near-fatal results in 2011. Shortly after birth, Williams noticed the characteristic symptoms of an embolism developing in her lungs and vocalized her concern to medical staff several times. It was only after repeated self-advocacy that doctors considered her pleas, and she was ultimately given life-saving treatment in part, if not entirely due to the power of her celebrity. Williams’ lived experience is used as an oxymoron to demonstrate the prejudiced, silenced, and lethal Black birthing experience, even among Black women with high social status. It is a testimony to the history of trauma as it relates to Black mothers’ need to convey emotion and expression as a prerequisite to humane treatment (McLemore, 2022). In a conversation between artists Andrea Chung and D’Yuanna Allen-Robb, both women agree that art, particularly in the form of photographs, can help shift consciousness and bring us closer to truthful narratives surrounding Black life and Black motherhood (Caruth, 2022). The archival process of depicting “truth” is muddied when one thinks of the harm subjected to the photographed replication of these mothers: removed from their original ownership, pawned, sexualized, and consumed. Zarse found that oftentimes, sexualized tags were directly linked to a racial description of the subjects. This leaves the archivist with the task of restoring dignity and safety to the contents within the archive.

In their chapter “What Gets Counted Counts”, Catherine D’Ignazio and Lauren Klein point out classification systems’ long history of projecting socially constructed ideas of gender and race onto systems employed by data researchers. Historically, race was first socially constructed during the transatlantic slave trade to categorize enslaved from enslaver. European ideas before this time demonstrate a religious and cultural categorization as it applies to race, but not a racial categorization based on the biological phenotypic presentation of the individual in question (D’Ignazio and Klein, 2020). As the eighteenth century progressed, classifications became more racist. As a result, librarians today may be tempted to do away with classification systems entirely; alas, data must be classified in order to use, and in order for archivists and researchers to highlight the vulnerability and predation of these images, we must consider their classification in the archive.

The women depicted in the photo collection have been robbed of their agency; we have no connections to their contact information, cannot garner how their images landed into the hands of these sellers, and do not know to what extent their consent was considered during the monetary profit of their pregnancy. This makes the archivists task of describing the collection particularly arduous. Library of Congress Subject Headings (LCSH) are infamous for their historical ties to the employment of patriarchy and racism in their classification systems, yet this remains the premier organizational protocol for metadata schemas. Although the Library of Congress has amended and eliminated some their more infamous patriarchal, homophobic, and racist subject headings, in the development of the metadata protocol for this specific photograph collection, I decided to use custom unstructured metadata in the descriptions of the photographs instead of LCSH standards. Recognizing that labeling unknown subjects with a socially constructed race could produce inaccuracies and racial bias paralleled in the original listings, I have opted to instead highlight the setting, background, pose, sense of community, and stage of pregnancy found in the images as a juxtaposition to their original listings.

Opacity and a Non-custodial Approach

The evergreen question of this archival process lies in the capture and upload of the images onto the digital archive, especially when one considers Chloé Meley’s declaration that “the implicit idea is that marginalized communities’ pain has to be portrayed, as if the telling of it wasn’t enough, as if additional, visual proof was required” (Meley, 2019). Sean Purcell undertakes similar questions in his scholarship in medical digital humanities. In his interactive photo essay “Dermographic Opacities”, Purcell employs the method of opacity to challenge the exploitative nature of knowledge systems, particularly in the field of medicine. Medical knowledge claims are rewarded at the expense of marginalized demographics, which can clearly be seen in the history of medicine- the theft of Indigenous people’s bones in anthropology and medicine (Redman, 2016); to the exploitation of Jewish people’s bodies by Nazi medical experimentation; to the theft and continued use of Henrietta Lacks’ cancer cells (Skloot, 2010); to the infamous Tuskegee experiment. From this culture, we have images marked with traceable biases. Photographs taken in the doctor’s office take advantage of subjects unable to say ‘no’ to their doctors, and once these subsequent photos were widely published in medical journals and textbooks, a system of reliance on this exploitation was set in place. Purcell follows Édouard Glissant’s “right to opacity” to undermine the totalitarian epistemics in biomedical science and promote the right to be forgotten, unknown, and extricated from systems that would perpetually harvest data from its prey. The images in the essay display varying levels of opacity imposed by Purcell that either partially or totally obscure the identity of the patient. In the obstruction, the image isolates the disease and abstracts its meaning from the subject. Once the identity of the subject has been obscured and separated from the medical condition, the image begins to break down systems of knowledge. The subject may appear to have been erased even more than the original depiction, but Purcell hopes that some dignity can be restored through this process. I see the practice of opacity as putting otherwise isolated photographs in conversation with one another, and this is particularly salient as Zarse and I amalgamate her purchased photographs into a singular archival body. As I apply this thinking to the photographic archive, I consider whether community archiving principles can also be imposed onto these photos.

In Patricia Franks’ “The Handbook of Archival Practice”, she defines community archives as, “grassroot archives created and stewarded by community members to preserve shared histories, experiences, and identities” (2021). Community archives often build foundational frameworks around decolonial and repatriated. In “Archives and Special Collections as Sites of Contestation”, Mary Kandiuk states that community archives were born out of the recognition that “[c]olonial practices have dominated recordkeeping, description, and spaces, and a culture of privilege and exclusion has prevailed over community access and engagement. There is strong interest in documenting historical injustices and reaching out to historically marginalized communities, thereby transforming interpretations and creating new narratives” (2020). As the archivist for this collection, I am to provide visitors with a more nuanced narrative to the collection than what might initially meet the eye.

The terms anti-custodial, non-custodial, or post-custodial are often elicited when describing community archives. These terms can be interpreted in a myriad of ways when applied to community archives, but I have historically employed these terms to classify that the materials that are gathered in the archive are owned by the community and its members, not universities or institutions. This sentiment of oppression and custody in the architectural structure of archives was introduced by Luciana Duranti as early as 1996. Assuming that communities will be owning their materials before, during, and after the production of the archive, it is critical to think about where the architecture of the archive, especially as it relates to where materials will be stored and who will have access to these materials. Although ownership of a singular object in the collection typically lies with a singular individual, Michelle Caswell and her team wrote the article “Imagining transformative spaces: the personal-political sites of community archives” based on how members of marginalized communities imagined the physical spaces of community archives. After talking with ten focus groups with a total of fifty-four participants, her team found that members of marginalized communities view community archives as symbols of representation, as a home-away-from-home, and as politically generative spaces.

In order to create these spaces of representation, home, and political generation, post-custodial archival theory proposes that archivists move past the typical procedure of incarceration that is employed by most institutions, wherein materials live in a box that is locked away from the public and only brought out of confinement upon the special request of the patron. Alternatively, post-custodialists propose that objects remain in the eyesight and possession of the original owner throughout and after their digital capture and archiving (Hicks-Alcaraz, 2022). During the collection process for digital community archives, photos are taken to capture the ephemeral objects that contributors bring to digitize and “donate” to the archive. However, in this case, the digital copy of the object is stored in the archive, not the physical original. By maintaining a non-custodial theory, contributors hold onto their possessions throughout the collection, watch the process, contribute the digitized objects and an oral history, and have full control over the accessibility and permissions of their donations, which can change at any time in the collection and post-production process. At the end of the digitization process, the object is permanently returned to the original owner. However, when one considers Zarse’s curated collection, wherein the objects have already been removed from their original ownership, pursuing return does not seem ethically sound, as this would require an increase in the digital circulation of the photographs in the hopes that a subject or relative would recognize and reclaim the object, which would further subject the photos to additional vulnerability. Harkening back to Purcell’s work, in order to reduce circulation and harm, I feel compelled as an archivist to consider imposing opacity onto the photographs that do not naturally obstruct the facial identity of the subject.

Archiving The Pregnant Image

After taking these considerations into mind, I began the archival process and tailored the custom-made dataset that I originally made for Zarse into a metadata template for the archive. This digital archive and website, The Pregnant Image, was created using CollectionBuilder, an open-source framework for creating digital collections that are metadata driven and powered by modern static web technology. In order for CollectionBuilder to run, the website developer must upload a metadata profile unique to the site itself. I intended for each photo, envelope, and ephemera included in Zarse’s purchased envelopes to be captured as a singular record, each of which correlates to a separate row in the dataset/metadata for each physical object in the collection. Because CollectionBuilder’s features are easily customizable and powered by metadata, the records will only populate when the corresponding metadata is uploaded to the template. However, an object does not necessarily need a digitized representation to populate the site, it only needs the metadata record. This suited the project’s decision to host an abstracted view of the archive without reproducing these vulnerable images and subjecting them to potential harm from the masses. I decided to upload scanned reproductions of photographs that naturally obstructed the facial features of the subject(s) but chose to keep images with identifiable subjects private and unavailable to site visitors.

On the back end of the site, the metadata profile contains customized fields to identify and describe Zarse’s research interests as well as ethical considerations. For example, Dublin Core metadata standards would define “creator” of a photograph as the individual who captured the shot with their camera. As this is mapped onto The Pregnant Image, every creator is listed as “unknown” to highlight the anonymity and exploitation of the original owners. Any handwritten script found on the back of the photographs is also included in the metadata profile, as this is the only remaining connection to creator/owner’s authentic interpretation of the photograph. To align with Zarse’s interests, I included fields to describe the subject’s stage of pregnancy, which utilizes a controlled vocabulary of “prenatal”, “labor”, “birth”, and “postnatal”, a description of the subject’s pose, whether the subject’s face is obstructed, the presence of support in the photograph, and the support’s implied role. Finally, I also included metadata fields that would identify the seller’s description of the photography and the seller’s racial description of the subject, both of which are included in an effort highlight the prejudice leaking into the archive. Visitors to The Pregnant Image can view the first ten envelopes of the collection by navigating to the “Browse” tab.

CollectionBuilder utilizes modern static site web technologies in the construction of the websites produced from their template. A static site exclusively hosts pre-written content that remains unchanged (unless manual edits are made by the website developer on the back end), meaning that all the site’s information is stored in files that are then projected onto a html page. These pages are consistent across visitor use, remaining fixed and therefore static across user experience. Static sites are simple to build, but they also employ minimal computing technology. Alex Gil advocates for scholars around the world—librarians, professors, students, cultural workers, and independent researchers to ask the question, “What do we need?” when engaging in website development. Gil hopes that this question will encourage the development team to pursue a minimalist approach to the project and leave behind energy consuming add-ons. This is particularly salient as one compares static vs. non-static, or dynamic, websites. Because dynamic sites are constantly re-generating code to form their user-facing pages, this requires a great deal more computing energy and memory in the building, maintenance of, and visitation to the site as opposed to static site technology. “Minimal computing shares these concerns, prodding a creative practice that seeks to reduce our impact while achieving our needs” (Gil, 2015). Therefore, minimal computing supports a more ecologically sound web development process by eliminating unnecessary expenditures of computing energy that exacerbate environmental impact.

The conversations surrounding The Pregnant Image are ongoing, and this is reflected in the “About” drop-down menu on the site, which details a timeline of meetings and engagement Zarse and I have coordinated since I undertook the archiving of the collection. “Data Ethnography” hosts an essay about the collection, and the “Technical Information” page hosts information about the metadata driving the digital archive. The site currently hosts ten envelopes from the archive, identifies the content, context, and users of the project, describes the Metadata Application Profile, discusses archival practices, and documents the collaborative nature of the project. The Pregnant Image tackles the paradox surrounding digital archiving and repatriation; if archivists attempt to reunite salvaged photos with their original owners, this will only perpetuate their mass circulation, making the materials vulnerable to harm by the very nature of widespread dissemination. In almost all collaborative meetings about the collection, discussions on maternal mortality rates in the United States and the recent reversal of Roe v. Wade have surfaced. More time needs to be devoted to the research and application of these topics into the greater conversation of The Pregnant Image. As this project continues to develop and describe more objects in the collection, it is my hope that these conversations tied to surveillance, privacy, and voyeurism will be embedded in the digital preservation of these photographs.