What Is EPUB? It’s a Website in a Zip File

Here is the fastest way to understand EPUB. Take any unprotected .epub file, rename the extension to .zip, and open it. Inside you will find a folder of HTML files, a stylesheet, some images, and a couple of small XML files that list the contents in order.

An ebook is a small website in a box. Once you know that, everything else about the format follows.

What Is Actually Inside

• One HTML file per chapter — the text itself, marked up as headings, paragraphs and lists.
• A stylesheet — fonts, spacing, indentation. The same CSS a web page uses.
• Images, including the cover.
• A manifest and spine — XML files listing every resource and the order chapters should be read in.
• A navigation document — the table of contents.

The whole thing is zipped, and the zip is renamed .epub. That is the entire format. It is an open standard maintained by the W3C, which is why it opens on hardware from a dozen different manufacturers.

Reflowable: The Thing That Makes It an Ebook

Because the content is HTML, it has no fixed pages. The text flows into whatever screen it finds, the way a web page does.

Change the font size and the text re-flows. Rotate the device and it re-flows. Read the same book on a phone and a tablet and the number of pages is different on each, because pages are not stored in the file — they are computed at the moment of display.

This is the format’s central advantage and it has one notable consequence: page numbers do not exist. There is nothing to number. This is why e-readers show a position or a percentage instead, and why citing an ebook is awkward in a way citing a printed book is not.

It also explains the division of labour against PDF: a PDF preserves a page that someone designed, an EPUB abandons the page so the reader can set the terms. Neither is better; they answer different questions.

Fixed-Layout EPUB

There is an exception, and it is worth knowing because it is often chosen by mistake.

Fixed-layout EPUB pins each page to exact coordinates — used for comics, picture books, cookbooks, anything where the arrangement carries meaning. It keeps the design intact and gives up everything reflow provided: the reader can no longer enlarge the text, and on a small screen the result is a tiny page that must be pinched and panned.

Use it only when the layout genuinely is the content. For a book of prose it is the wrong choice, and readers with poor eyesight will be the first to say so.

EPUB 3, and Why It Matters for Chinese

EPUB 3 is the current generation, and it added things that turned out to matter more than they first looked:

• Vertical writing and right-to-left page progression — which is what makes a properly typeset Chinese or Japanese ebook possible at all.
• Ruby annotations — the small pronunciation characters beside a word, essential in Japanese and used for less common Chinese characters.
• Embedded fonts, audio, video, and MathML for equations.
• Media overlays, synchronising narration with the text.

If an ebook has to handle CJK text seriously, EPUB 3 is the floor, not an upgrade.

DRM Is Not Part of EPUB

This confuses a great many people, and the distinction is worth being precise about.

EPUB itself contains no copy protection. DRM is a wrapper that a store adds on top: the file is encrypted, and only an approved application holding the right key can open it. The format underneath is unchanged.

Practically, this means a purchased ebook may refuse to open in the reader of your choice — not because EPUB is restrictive, but because that particular copy is locked. A DRM-free EPUB, the kind sold directly by many publishers, opens anywhere and can be renamed to .zip as described above. A locked one cannot.

Where It Opens

Essentially everywhere: Apple Books, Google Play Books, Kobo, Nook, and free readers on every desktop platform. Browsers can read it through extensions.

Kindle was the long-standing exception, having pushed its own formats for years. Amazon now accepts EPUB for documents sent to a device and converts it internally — so sending an EPUB to a Kindle works, while the file on the device is no longer an EPUB.

Four Ways an EPUB Goes Wrong

• Images at print resolution. A photograph sized for paper can be several megabytes; a dozen of them make an ebook that is slow to open and expensive to deliver. Size images for screens.
• Fonts not embedded. The book looks right on your machine and substitutes something else everywhere the font is missing — most visibly with Chinese, where the fallback may not contain every character.
• A table of contents that does not match. The navigation document is a separate file from the text; edit chapters without updating it and the contents list points at the wrong places.
• Layout forced with hard breaks and spaces. Since the text reflows, anything positioned by manual spacing will land somewhere different on every screen. Use styles, not spacing.

Quick Reference

• A zip archive of HTML, CSS and images — rename to .zip to look inside
• Reflowable: no fixed pages, which is why page numbers do not exist
• Fixed-layout EPUB keeps the page and gives up resizable text — comics only
• EPUB 3 adds vertical writing, ruby, embedded fonts and media
• DRM is a wrapper added by stores, not part of the format
• Opens everywhere; Kindle accepts it and converts on arrival
• Embed fonts, size images for screens, keep the TOC in sync