Should ebook text be justified or ragged right?
Either works when each line holds enough words and the software hyphenates, because the extra space is then shared out in small amounts. On a narrow page at a large font size, ragged right reads better: there are fewer gaps to share the space between, so each gap grows. Amazon's layout engine allows for this: its help page for publishers says a Kindle switches to left-aligned text by itself when the font is large or the line is short.
Where the setting lives
On a Kobo, tap the middle of the page to open the Reading Menu, tap the Text icon, and pick an option beside "Justification" (Kobo help, November 2026). In the Kobo Books app for phones the setting is called "Alignment" (Kobo help). A Kindle book made for Amazon's newer layout engine opens fully justified, and the same publishing help says readers can change that to left aligned (ragged right). In April 2025 Amazon wrote that for some readers left-aligned text, with wider spacing and margins, can reduce eye strain.
Why printed books justify
Many readers take a straight right edge as the mark of a proper book. Matthew Butterick, a typographer, calls justification a matter of personal preference, not a sign of professional work, in Practical Typography. He notes that books tend to justify while newspapers and magazines mix both, yet he left-aligns his own web pages and word-processor documents, because the justification in browsers and word processors is crude next to a page-layout program.
Early printers also had more ways to fill a line. In Gutenberg's 42-line Bible, as Johannes Ammon describes it, the compositors had about 290 different glyphs to choose from, including lowercase letters of more than one width and a heavy use of Latin abbreviations, and they let hyphens hang slightly into the margin. A printer fits each line once, for one typeface at one size. An e-reader has to do it again every time you change the font, the size or the margins.
What the reading studies found
Studies from 1986 and 1991 disagree about whether justified text is slower to read. Justified text was slower in two studies that Stanley Trollip and Gregory Sales ran in 1986, when computers often justified a line by inserting extra spaces between words. Comprehension didn't differ.
Five years later, Rebecca Jubis and Paul Lee tested 96 university students on a CRT screen with four layouts. Justification made no difference to reading time. Finding typographical errors was fastest with "even-gap" justification, where small spaces were spread proportionally between the words, and slowest with ragged text whose word gaps varied. Their explanation was that a reader scans more slowly when unsure where the next word starts.
Rivers and readers with cognitive disabilities
Rivers, the channels of white that form when gaps line up down a page, are often treated as a cosmetic flaw. The web's accessibility standard takes them more seriously. The W3C's explanation of WCAG success criterion 1.4.8 says some people with cognitive disabilities find justified text hard to read, that rivers can make reading difficult or even impossible, and that squeezed lines make word boundaries hard to find. Among other things, the criterion asks for a way to get text that isn't justified. It also caps the length of a line, which the post on line length takes up.
Hyphenation and the language of the book
Butterick's rule is that justified text needs hyphenation, or the word gaps become huge. The software that does it misses breaks on purpose. Frank Liang's 1983 Stanford thesis, Word Hy-phen-a-tion by Com-put-er, produced the pattern method that is standard in TeX and has since been adapted to many languages. His English set of about 4,500 patterns found 89% of the permitted breaks in a pocket dictionary's word list, and put virtually none in the wrong place. The missing 11% was a choice: the last breaks proved expensive to capture, and Liang judged a missed break safer than a wrong one. A few can't be found from the spelling at all, such as "re-cord" the verb beside "rec-ord" the noun, and the dictionaries don't always agree either: the thesis shows Webster's and the American Heritage breaking "independent" differently.
Patterns are per language, so the software has to know the language first. The CSS Text standard, which EPUB styling builds on, lets a reading system hyphenate automatically only text whose language is declared. Readwyn works the same way: it hyphenates a book whose language is English with the Knuth and Liang patterns of TeX, and breaks a book in any other language, or one with no language set, only between words.
Not every script stretches its word spaces. The W3C's note on approaches to full justification explains that Arabic can lengthen the joins between letters (kashida), that Chinese and Japanese, which have no word spaces, adjust the space around punctuation and between characters, and that Thai breaks lines at word boundaries that aren't marked by any space.