F-pattern reading: what eye-tracking research reveals about how we scan text

In 2006, Jakob Nielsen published what became one of the most-cited findings in web usability research. His team at the Nielsen Norman Group tracked the eye movements of 232 participants reading thousands of web pages. The aggregate gaze plots produced a consistent shape: an F.

Participants read the first line or two in full, covering the width of the text block. Then they made a shorter horizontal sweep on the next line or paragraph. Then shorter still. Attention concentrated increasingly toward the left margin as the eye moved down the page — producing the two horizontal bars of an F and a vertical spine running down the left edge.

This was not a strategy readers had chosen. It emerged from aggregated gaze data as an unconscious output of a brain trying to extract maximum information in minimum time.

What the F-pattern actually shows

The F-pattern is a visualisation of information-triage behaviour under uncertainty. When a reader does not know whether a document is relevant or worth reading in full, the brain defaults to a scanning heuristic: sweep the beginning of each line, gather structural signals, and progressively reduce the horizontal reach as confidence in the document's utility either grows or falls.

Nielsen's 2006 heat maps showed three characteristic components:

1. First horizontal movement: readers read across the top of the content, covering most of the first line — absorbing the document's opening claim or headline continuation.
2. Second horizontal movement: a shorter sweep lower on the page, typically covering the beginning of a subsequent paragraph or a subheading area.
3. Vertical movement: a slow scan down the left side of the content, picking up the first words of paragraphs.

The pattern gets its F shape from components 1 and 2 forming horizontal bars, with component 3 forming the vertical spine.

The five scanning variants: Pernice 2017

In 2017, Kara Pernice extended the original research at the Nielsen Norman Group, reporting on substantially more participants and a broader range of page types. Her study identified five distinct scanning patterns — not a single universal F, but a family of related strategies.

F-pattern: The classic form. Dominant on dense text pages without strong visual hierarchy. Most common when readers approach a page with low commitment or uncertain relevance.

Z-pattern: The eye sweeps across the top of the page, then drops diagonally to the bottom-left corner before sweeping right again — tracing a Z or N shape. Common on pages with minimal text and image-heavy layouts, and on pages the reader is scanning for a specific element.

Layer-cake pattern: The eye reads headings and subheadings but skips the body text between them entirely, landing on each heading like rungs on a ladder. Emerges when document structure is strong and the reader is using headings as a navigation system rather than reading the content.

Spotted pattern: The eye fixates directly on specific elements — numbers, highlighted text, proper names, or list items — without following any systematic path. Characteristic of readers who have a very specific information target in mind and are executing a high-precision scan.

Commitment pattern: Near-full reading of every line, with consistent horizontal sweeps across the full text width. Emerges when content is judged highly relevant, personally important, or unfamiliar enough to require careful processing.

The critical insight from Pernice's taxonomy: readers do not have one scanning behaviour. They have a repertoire, and they shift between modes based on content structure, perceived relevance, and reading goal. The F-pattern is the default triage mode, not the only mode.

The perceptual span: why partial reading yields more than you expect

Eye-tracking research reveals that readers do not see only what they fixate on. Rayner (1998, DOI: 10.1037/0033-2909.124.3.372) established that during each fixation, the visual system extracts useful information from approximately 14–15 letter spaces to the right of the fixation point — equivalent to 2–4 words of context. Letter shapes and word lengths are processed further into the periphery, up to about 20 character spaces.

This perceptual span means that when a reader's eye lands at the beginning of a line, words 2 through 5 on that line are simultaneously registered in peripheral vision. The F-pattern's horizontal sweeps therefore capture considerably more than the directly fixated words. And when the eye moves down the left spine, adjacent words on each line are picked up peripherally without a direct fixation.

This is why F-pattern reading does not produce comprehension as poor as naive word-counting would suggest: each fixation pulls in a cone of context. The technique's efficiency comes from the brain's parallel processing architecture — foveal processing for the fixated word, parafoveal processing for the next 2–4 words, peripheral processing for word shapes beyond that.

What the F-pattern means for how you should read

Understanding the F-pattern gives you two practical tools.

When reading deliberately: Use the pattern as a foundation for diagonal reading. Lean into the natural staircase movement of your eye rather than fighting it. Read the first sentence of each paragraph in full (the horizontal sweep), then drop to the next paragraph's first sentence — rather than reading the full paragraph first. This converts an unconscious triage habit into a controlled comprehension strategy.

When skimming: Deliberately amplify the F-pattern's vertical spine component. Spend more time on the left-margin scan and use the spotted pattern for numbers, proper names, and key terms. Setting a clear reading goal before you start — "I want to know the main finding and the methodology" — activates the spotted pattern for relevant content words, improving scan precision. See metacognitive reading for how goal-setting affects fixation patterns.

When writing: Front-load key information. The F-pattern's first horizontal sweep covers the opening line more fully than anything else on the page. Paragraph-opening sentences receive the most attention; closing sentences receive the least. A reader scanning with an F-pattern will absorb your first sentence reliably and your last sentence rarely.

From F-pattern to diagonal technique

The diagonal reading technique is a systematised and trained version of the F-pattern. Where the F-pattern emerges spontaneously and inconsistently — varying in depth and coverage depending on perceived relevance and reading fatigue — diagonal reading applies deliberate path settings (word density and step angle) to control coverage and pace.

The diagonal path traces the same general trajectory as the F-pattern's vertical spine, but adds controlled fixation points at set intervals rather than relying on the eye to choose landing spots ad hoc. This reduces the cognitive load of deciding where to look, freeing working memory for comprehension rather than navigation.

Try it now: The Diagonal Reader tool lets you paste any text and visualise a diagonal scan path with adjustable density and angle. The animated guide mode trains the eye movement pattern until it becomes automatic. Free, no account required.

Learn the full method: The Diagonal Reading course covers the F-pattern evidence in Lesson 1, perceptual span science in Lesson 2, and the full diagonal technique training in Lessons 3–4. Six evidence-based lessons, free, no account required.

Further reading in this series

• What is diagonal reading? The evidence-based guide to structured skimming
• How to skim read effectively without losing comprehension
• The SQ3R method: why surveying before you read improves comprehension
• Metacognitive reading: why knowing your goal makes you faster
• Reading triage: how to decide what to read fully, skim, or skip

References

• Nielsen, J. (2006). F-Shaped Pattern for Reading Web Content. Nielsen Norman Group. https://www.nngroup.com/articles/f-shaped-pattern-reading-web-content-discovered/
• Pernice, K. (2017). Text Scanning Patterns: Eyetracking Evidence. Nielsen Norman Group. https://www.nngroup.com/articles/text-scanning-patterns-eyetracking/
• Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research. Psychological Bulletin, 124(3), 372–422. https://doi.org/10.1037/0033-2909.124.3.372
• Rayner, K., Schotter, E. R., Masson, M. E. J., Potter, M. C., & Treiman, R. (2016). So much to read, so little time. Psychological Science in the Public Interest, 17(1), 4–34.
• Carver, R. P. (1990). Reading Rate: A Review of Research and Theory. Academic Press.
• Klimovich, M., Tiffin-Richards, S. P., & Richter, T. (2023). Does speed-reading training work, and if so, why? Journal of Research in Reading, 46(2), 123–142. https://doi.org/10.1111/1467-9817.12417