At a glance

A single curator took Softness Was Always Ours from a one-page proposal to a nine-panel gallery installation and companion digital exhibit in under three months, then rebuilt it in 2026 around 35 library collections. The exhibit won two 2026 ACRL/RBMS Leab Awards.

The challenge

The exhibit set out to counter a familiar archival framing of Black women through struggle alone, and to show softness as something Black women have always claimed. Its stated aim: center grace, care, joy, and beauty in the everyday lives of Black women and girls across generations.

The constraints shaped every decision:

• One curator, alongside a full processing workload. The exhibit was built in parallel with a 197-box archival processing project.
• A compressed schedule. The proposal planned a six-week path from selection to launch.
• A modest budget. Printing, mounting, and programming had to be priced and justified line by line, with expenses tracked for grant reporting.
• Two audiences at once. The gallery had to work for walk-in visitors, and the digital exhibit for remote researchers and the public.

Planning and concept

The exhibit began as a one-page proposal that defined the title, themes, objectives, audience, and a week-by-week timeline, and was approved within days of submission.

• Proposal. Four founding themes (Adornment & Opulence, Softness in Strength, Joy & Sisterhood, Resilience in the Everyday), four objectives, a target audience of researchers, students, fashion historians, and the public, and a six-week schedule.
• Mood board. A visual board of comparable imagery let decision-makers see the intended look before any archival item was pulled.
• Color palette. Five candidate palettes were developed, each with a main, accent, background, and highlight color. Rose & Champagne was chosen for its vintage elegance and approachable warmth.
• Curatorial statement. One core statement was adapted into versions for the entrance panel, the display case, a press release, LinkedIn, and Facebook and Instagram, so every channel told the same story in its own register.
• Themes that grew. As research surfaced more material, the four founding themes became seven: Wedding Bells, Softness Swims, and Softness in Print joined the original four, plus a timeline.

Production management

A phase-based production checklist tracked every task by status, so the exhibit moved from approval to installation in about twelve weeks.

1. Planning and content development: image selection, condition checks, curatorial narrative, theme groupings, captions, and quotations from Black women writers.
2. Digitization and file prep: scanning, image editing, high-resolution print files, and compressed web versions.
3. Design and fabrication: panel templates, caption cards, print sizes, vendor pricing, and budget confirmation.
4. Logistics and installation: gallery scheduling, floor plan, facilities support, installation, and multimedia testing.
5. Outreach and programming: programming partners, press language, social media teasers, staff notice, and soft opening.
6. Ongoing: expense tracking for foundation and grant reporting.

Selection under real conditions. An initial list of 456 candidate images narrowed to 308 that were available and intact, then 248 pulled for review, and finally about 100 for the digital exhibit.

Vendor sourcing and budgeting. Five vendors were contacted for panels and pop-up banners. For the 2026 reprint, every one of 66 sub-panels was measured and priced by size, and two printers’ quotes were compared: about $253 for prints alone versus $430 with a 40 × 54-inch title wall and delivery.

Hand-offs that held. Print-ready files went to the printer in panel-specific folders, 118 images in all, so nothing was misplaced between design and production.

Gallery and digital design

The gallery was planned wall by wall as an emotional arc, so visitors move from innocence to joy to stillness before they leave.

1. Entry wall: title panel, curatorial statement, and a single portrait that sets the tone.
2. Long wall: a progression from childhood to adolescence to womanhood, with a large anchor image every four to five feet.
3. Far wall: couples, dancing, and community joy as the visual high point.
4. Facing wall: solo portraits of stillness and elegance, paired with quotations on rest and care, to slow visitors down.
5. Exit wall: a reflection station with QR codes.

Design standards. Anchor images at 16 × 20 or 18 × 24 inches, supporting images at 11 × 14 or 12 × 18, captions at letter or tabloid size, and image centers at the 57- to 60-inch gallery eye line. Walls stayed neutral; soft fabric accents behind the statement and the most evocative grouping carried the theme without permanent changes.

QR reflection station. Three codes linked the room to the web: the digital exhibit, a donation page for the library’s foundation, and the visitor feedback form. Each was tested before printing and printed at 300 dpi or higher.

Digital exhibit. Built and customized in Omeka S with HTML, CSS, and JavaScript: theme pages, a timeline, custom tile cards, extended captions, photo pages that open in their own tabs, and a companion playlist with full citations.

Community engagement and programming

The opening was staged over three days for three audiences: a private catered reception on September 16, 2025, a press day on September 17, and a public reception on September 18.

Targeted outreach. Invitations went to community organizations whose members the exhibit portrays, including Win With Black Women, metro Atlanta chapters of the Divine Nine, The Links, Jack and Jill of America, Top Ladies of Distinction, Iota Phi Lambda Sorority, and Sigma Pi Phi Fraternity. Listings went to local media and partner networks, and local arts press attended the opening.

A reading list tied to the walls. More than 120 titles were evaluated across poetry, history, fiction, self-care, and beauty, each mapped to an exhibit theme and checked against library holdings. Thirty-nine were chosen for the display shelves, supported by $500 in collection development purchases that outlast the exhibit.

Programming built for collaboration. Six program formats were proposed for the exhibit’s six-month run:

• A community panel or artist talk on Black femininity, fashion, and archival photography
• An intergenerational storytelling circle on womanhood and beauty
• Guided walk-throughs for schools, women’s groups, and partners
• A style-themed film screening or fashion retrospective
• A workshop on the cultural significance of Black hair
• A conversation on the historical policing of Black women’s appearance

Shared credit. Branded postcards, brochures, banners, and social media graphics were produced with the library’s graphic designer, and colleagues contributed artwork and display support.

Evaluation

A six-question visitor feedback form, linked from the exit wall, measured emotional impact as well as satisfaction, because the exhibit’s goal was a change in perception.

The form was designed to work on paper cards or an online form, so it could serve gallery visitors and remote viewers alike.

The 2026 revamp

In early 2026, the exhibit was rebuilt, physical and digital, around 149 items drawn from 35 of the library’s collections, without closing the gallery’s story or losing its themes.

The first version drew largely on a single photography collection. The revamp widened the lens to studio portraits, family papers, sorority and club records, scrapbooks, magazines, and newspaper clippings, spanning the 1880s through the 1980s.

How it was done.

1. Preserve the original. The first digital exhibit was archived before any change was made.
2. Map replacements to themes. A tracking workbook recorded every candidate item by theme, decade, collection, container, and sub-panel, with columns for digitization and upload status.
3. Digitize and print. Items were digitized in batches, and 67 replacement images were printed to recover seven gallery panels, sized to each sub-panel’s measured dimensions.
4. Rebuild the digital exhibit. Seventeen Omeka pages were re-engineered in HTML, CSS, and JavaScript, with new navigation, custom theme tile cards, a timeline, and extended captions for 141 images.

The six photographic themes each kept between 10 items (Softness Swims) and 38 (Resilience in the Everyday).

Results and recognition

The exhibit received two 2026 Leab Awards from the Association of College & Research Libraries’ Rare Books and Manuscripts Section, the first time the award’s recipients were chosen by popular vote of members.

• Excellence in Content and Innovation in Audience Engagement, digital catalog category, alongside recipients such as the Morgan Library & Museum and the University of Toronto (announcement).
• The curator presented the method nationally in a Society of American Archivists E2 Hour session, From Negatives to Narrative: Building a Digital Exhibit from Archival Collections (April 2026), which drew record registration for the series.
• The digital exhibit remains live and freely available: Softness Was Always Ours.

Transferable practices

Five habits made a one-curator exhibit possible, and each transfers to any library or archive with a small team.

1. Win approval with a picture. A one-page proposal plus a mood board let leaders say yes quickly, before any archival item was handled.
2. Write once, adapt everywhere. One curatorial statement, rewritten for each panel and channel, kept the message consistent from the gallery wall to social media.
3. Run production from a phased checklist. Grouping tasks by phase and status made gaps visible early and gave supervisors a simple progress view.
4. Price at the smallest unit. Measuring and pricing every sub-panel turned printing from a guess into a defensible budget and a fair vendor comparison.
5. Design for change. Tracking every item by theme, collection, and panel position meant the exhibit could be rebuilt from new collections in weeks, not months.

Project snapshot

In spring 2025, I designed and built a web-based study application for the Certified Archivist (CA) examination: a bank of more than 1,000 source-cited questions organized by the eight ACA exam domains, delivered through an adaptive quiz interface with learning and test modes, filters, and domain-level analytics.

The challenge

The CA exam tests eight practice domains drawn from a reading list of roughly 50 books, articles, and standards, and ACA says questions come from that literature without a one-to-one link to any single reading. Candidates have the study guide and the list, but no structured way to practice against the full body of literature or to see which domains need the most work.

I was preparing for the exam myself in spring 2025, alongside full-time archival work. Generic flashcards and loose notes would not show me where I was weak, and they would not tie a missed question back to the source that explains it.

Approach

I built from the exam’s own framework outward rather than inventing a curriculum. The ACA Study Guide’s eight domains became the top-level structure, and the ACA Reading List became the source pool every question had to trace back to.

Three principles guided the work:

• Source-grounded: every question carries an explanation and a citation to the reading it came from, down to the page where possible.
• Structured, not prose: questions are records with consistent fields, so an application can filter, randomize, and score them.
• Diagnostic: results break down by domain, so the tool tells a candidate where to study next, not just how many they got right.

I used ChatGPT as a drafting and engineering partner: extracting concepts from readings, drafting questions and distractors, restructuring data, and writing and debugging code. I kept the subject-matter judgment, verification, and design decisions.

Process

The core build took about one week, from a first PowerPoint prototype on April 30, 2025 to a specified, coded app with three question formats by May 7. The following four weeks went into source coverage and study supplements.

Data and metadata architecture

Each question is a record, not a paragraph. Drafting started in Word documents organized by source, moved into spreadsheets and CSV, and ended as JSON the app could load.

This is ordinary archival practice applied to assessment: consistent description, controlled values, and provenance for every item. The same fields that make a finding aid searchable make a question bank filterable by domain, source, and format.

Interface and learning design

The app had three screens (start, quiz, and results) and two modes that serve different stages of study. I specified the interface in a written design document on May 6, 2025.

A side menu let a candidate build a session by domain (one, several, or all), question type, and source. Two filters made it adaptive: Unasked Questions Only and previously missed questions. Navigation included Previous, Next, Submit, a Flag for Review checkbox, and a progress bar reading “5 of 60.”

The results screen reported an overall percentage and a percentage for each domain, so the next session could target the weakest area.

Quality assurance and source coverage

AI drafted quickly, so verification was a separate, deliberate step. On May 5, 2025, I ran a verified-answers pass across all eight domains, checking each correct answer and explanation against the literature before it went into the app.

I managed coverage with three tracking tools:

• Targets by source: planned counts per reading and format, for example 50 multiple-choice items from Harvey’s Preservation Management Handbook and 40 from Ritzenthaler.
• Running totals by domain: finished questions by domain and format, separate from the targets.
• Source coverage matrix: every reading-list item mapped to the domains it supports, with format (Kindle, print, PDF), whether I had it on hand, and whether I had read it.

The matrix showed gaps that question counts hid. In its first version, Cultural Competency drew on a single book, and the later version maps the reading list’s articles on radical empathy, community archives, and reparative practice to that domain.

Outcomes

I used the app to prepare for the exam and earned the Certified Archivist credential in 2025.

• Question bank: planning targets set on May 2, 2025 called for 1,484 items (1,274 multiple choice, 138 true/false, 72 drag-and-drop). The finished bank exceeded 1,000 questions.
• Working product: a coded web app with two study modes, multi-domain filtering, adaptive review, and per-domain scoring.
• Study ecosystem: study plans, a domain confidence tracker, key-concept summaries, flashcards, audio overviews, and video lessons built around the same domain structure.
• Reusable assets: the question data, design specification, and coverage matrix survived the loss of the app and are the basis for a rebuild.

Lessons learned

The most important lesson came from losing the app. After errors in the Replit project, I deleted it, and the code went with it. The question bank, design specification, and coverage tools survived because they lived in separate storage, apart from the code.

• Keep content and code apart. Content stored as structured data outlives any one platform. Code needs version control and off-platform backups.
• Verification is not optional. AI made drafting fast; a dedicated answer-verification pass made the questions trustworthy.
• Track coverage, not only counts. A high question count can still leave a domain thin.
• Write the spec before the code. A plain-language interface document kept a fast, AI-assisted build consistent.
• Document as you go. Dated project files made this reconstruction possible a year later.

Skills demonstrated and next steps

Next, I plan to rebuild the app as a subscription product, with the code under version control and the question bank maintained as a standalone, versioned dataset.

White paper

• Source-Grounded and Human-Verified: An Archival Framework for AI-Assisted Assessment Design

  A six-stage framework that treats a question bank as an archival collection, with provenance, description, and appraisal as the controls for AI-drafted items. Includes recommendations for practitioners and certifying bodies. 6 pages, October 2026.

  PDF