Why Accessible PDFs Remain Essential in Academic Publishing

EPUB has been called the future of academic publishing for the better part of a decade. Reflowable text, adjustable font sizes, built-in accessibility affordances: on paper, it should have pushed PDF into retirement by now. It hasn't. Walk into any university library, open any journal database, download any paper for peer review, and PDF is still what shows up. Understanding why matters more than repeating the EPUB prediction one more year.

Fixed layout is a feature here, not a limitation

The case against PDF usually starts with its rigidity: text that doesn't reflow, layouts that don't adapt to screen size. For a novel, that's a real drawback. For a chemistry paper with a five-line equation, a data table with forty columns, or a figure whose caption depends on exact spatial position relative to the image, rigidity is the point. Reflowable formats struggle with content where position carries meaning. A structural diagram that shifts relative to its labels stops making sense. An equation that reflows across a line break becomes unreadable. PDF holds the layout exactly as the author and typesetter built it, which is precisely what dense STEM content requires.

Citation infrastructure runs on fixed pages

Scholarly citation depends on stable page numbers. "See page 47" only means something if page 47 is the same page for every reader, in every format, indefinitely. Reflowable EPUB breaks that assumption by design, since the same content lands on a different "page" depending on font size and device. Academic publishing hasn't found a citation convention that survives that instability at scale, which is a large part of why journals, dissertations, and reference works still anchor to PDF as the citable version of record.

Institutional workflows were built around it

Interlibrary loan, course reserve systems, and print-on-demand pipelines all assume a PDF at the center. Researchers still print papers to annotate by hand. Peer review platforms exchange PDF drafts as the default working format. None of this is nostalgia. It's infrastructure that took decades to build and that a large share of the academic world still depends on daily. Replacing it isn't a matter of publisher preference. It requires the entire ecosystem around research and instruction to move together, and that hasn't happened yet.

Archival systems are built on PDF/A for a reason

Long-term preservation systems like Portico and CLOCKSS standardized around PDF/A because self-contained, fully embedded files survive format migrations more predictably than formats that depend on external rendering engines and evolving reflow standards. A publisher's entire backlist commitment to long-term access runs through PDF as the archival format of record. That commitment doesn't disappear because a newer format exists.

None of this excuses an inaccessible PDF

Here's where the conversation usually goes wrong. Publishers hear "PDF isn't going anywhere" and treat that as permission to leave PDF accessibility as an afterthought, on the assumption that EPUB will carry the accessibility burden instead. That reasoning breaks down the moment a reader who needs a screen reader downloads the PDF, because it's still the version most institutions supply by default.

A genuinely accessible PDF has a tagged structure tree that mirrors the document's actual organization, not just its visual layout. It has reading order that a screen reader follows correctly, alt text on every image and figure that isn't purely decorative, a declared document language, and navigable headings and bookmarks. None of that happens automatically when a manuscript gets typeset. It has to be built in deliberately, and it's exactly the part of PDF production that gets skipped under deadline pressure.

STEM content is where this gets hardest, and where it matters most

Equations, chemical structures, and multi-layer data tables are the content types most likely to arrive at final PDF with no accessible tagging at all, because they're the hardest to tag correctly. An equation rendered as a flattened image tells a screen reader nothing. A forty-column data table with no header associations reads as an incomprehensible string of numbers. These aren't edge cases in academic publishing. They're the core content of entire disciplines, and skipping their accessibility work means locking out exactly the students and researchers who most need to engage with technical material on equal footing.

This isn't a PDF-versus-EPUB decision

The honest framing isn't choosing one format and abandoning the other. Most academic and STEM publishing programs need both: an accessible EPUB for readers who benefit from reflow and adjustable presentation, and an accessible PDF for the citation stability, institutional workflows, and archival requirements that still run through it. Treating PDF as the format publishers only owe half an effort to, because EPUB exists as a backup, leaves a large share of an institution's readers without a usable copy of the exact document their library actually distributes.

Backlist and frontlist PDFs both need the same deliberate work: structural tagging, verified reading order, meaningful alt text, and PDF/UA conformance checked against real assistive technology, not just an automated scan. That's the specific gap S4Carlisle's PDF remediation service closes, applied consistently across a publisher's full catalogue rather than title by title as complaints surface.

PDF isn't a legacy format publishers are stuck maintaining until EPUB finally takes over. It's the format academic publishing still runs on, for reasons that have nothing to do with inertia. Making it accessible isn't optional work bolted onto a format on its way out. It's core production work for the format that's still doing most of the job.

S4Carlisle is Benetech GCA certified and remediates PDFs to full accessibility conformance through the NINJA PDF Accelerator (s4carlisle.com/ninja-ai-suite/pdf accelerator) and the NINJA AI Ecosystem. Get in touch to see what it would take to bring your PDF catalogue up to standard.