WebBuilding

Foundations

Hypertext

Summary

Hypertext is text that contains links to other text, so a reader can jump between passages instead of reading in a fixed order. Vannevar Bush sketched the idea in his 1945 memex essay, Ted Nelson introduced the word in 1965, and Douglas Engelbart's NLS and Apple's HyperCard (1987) put it to work. Tim Berners-Lee's March 1989 proposal carried it across a network, and it became the World Wide Web.

Web Building Encyclopedia · Foundations · Last reviewed October 22, 2026 · 11 min read

What hypertext is

Hypertext is text that contains links to other text. Instead of reading from the first page to the last, a reader can jump from one passage to another by following a link, and the writer decides where those links lead.

The idea is older than the personal computer, but it only became everyday reading with the World Wide Web. When Tim Berners-Lee proposed the web at CERN in March 1989, he defined hypertext as “Human-readable information linked together in an unconstrained way.”1 He used the word hypermedia for the same thing when it is not bound to text, such as linked pictures or sound.1

A few terms come up again and again:

  • Node. One unit of content, such as a note, a card or a page. Berners-Lee’s 1989 proposal drew hypertext as circles and arrows and called the circles nodes.1
  • Link. The connection from one node to another, the arrow in that drawing.1
  • Anchor. The words or image a link starts from, or the spot it lands on. A 1992 page of the CERN project explained that the text inside an anchor tag can be the start of a link, its destination, or both.2
  • Browser. The program a reader uses to move through linked documents. The 1990 web proposal gave that name to any program that provides access to the world of hypertext.3

The word survives in the names of the web’s two core technologies. HTML stands for HyperText Markup Language, and HTTP for Hypertext Transfer Protocol, which the current HTTP standard describes as a protocol for distributed, collaborative hypertext information systems.4

Vannevar Bush and the memex (1945)

The history usually starts with an essay. In July 1945 The Atlantic Monthly published “As We May Think” by Vannevar Bush, whom the magazine’s editor introduced as Director of the Office of Scientific Research and Development.5

Bush blamed much of the trouble of finding information on the artificial way it was indexed: filed alphabetically or numerically, and found by tracing it down from subclass to subclass.5 “The human mind does not work that way. It operates by association,” he wrote.5

His answer was a machine he called the memex, which he presented as a future device for individual use, a kind of mechanized private file and library. It would store a person’s books, records and communications, and he described it as a desk that otherwise looked ordinary.5 W3C‘s history of the web sums it up as a photo-electrical-mechanical device that could link documents stored on microfiche.6

The heart of the design was what Bush called associative indexing, a way for any item to call up another automatically. A user would join two items into a trail, give the trail a name, enter it in a code book and tap it out on a keyboard. Later, tapping a button would bring back the item joined to the one on screen.5 Bush predicted new forms of encyclopedias that would arrive with trails of associations already running through them.5

Bush wrote about the memex as a device of the future, and the essay was read by the people who later built hypertext. Douglas Engelbart’s 1962 research report includes a section titled “What Vannevar Bush proposed in 1945.”7 Berners-Lee has written that he first came across the essay in documentation for a later memex project at Digital Equipment Corporation, and he called it a great paper, though he said there was no direct line from Bush, Engelbart or Nelson to the web.8

Ted Nelson, the word “hypertext” and Project Xanadu

The word itself comes from Ted Nelson. His Project Xanadu describes itself as founded in 1960 and as the original hypertext project.9 Nelson dates the start of his work to 1960 as well.10

The paper usually credited with introducing the term is “A File Structure for the Complex, the Changing and the Indeterminate,” which Nelson, then at Vassar College, presented at the 20th national conference of the Association for Computing Machinery (ACM) in a session on August 24, 1965.11 In it he wrote, “Let me introduce the word ‘hypertext’ to mean a body of written or pictorial material,” connected in ways so complex that it could not conveniently be presented on paper.11 The paper also proposed a file structure he called the ELF, or Evolutionary List File, based on an arrangement he called zippered lists.11

Nelson has pointed to a second 1965 paper, “The Hypertext,” as his actual first publication using the word, and believes it was presented at the 1965 conference of FID, the international documentation federation. It proposes hypertext as a new medium and calls it a generic term for texts that, because of their structure, need automatic handling and display devices.12

Sources differ on the date. Nelson’s own chronology says the term was introduced in 1965, and W3C’s history places it in a 1965 paper for the ACM.106 Berners-Lee’s 1989 proposal, by contrast, says Nelson coined the term in the 1950s.1

Xanadu aimed well beyond links. Nelson describes its key idea, transclusion, as showing the same content, knowably and visibly, in more than one place, which he distinguishes from making copies.10 The project’s site proposes parallel pages that are visibly connected, and a license it calls transcopyright, which permits reuse with automatic credit to the author.9

The software took decades. Nelson’s chronology says he believes John Walker at Autodesk built the first working interactive version of its side-by-side “transpointing windows” in 1988.10 On August 23, 1999, Udanax.com and Project Xanadu released two Xanadu systems, Udanax Green and Udanax Gold, as open source at the O’Reilly Open Source Convention.13 In 2014 Engadget reported that a Xanadu document viewer, which shows quoted material beside the document that quotes it, had been shown at a Chapman University event that April.14

Berners-Lee says he knew of Nelson by name before reading his work, which he came across while working on the web, probably between March 1989 and September 1990. He later ordered Nelson’s book Literary Machines and paid Nelson for it in person, in Sausalito, California, in August 1992.8

Engelbart’s NLS and the first working systems

While Nelson named hypertext, Douglas Engelbart’s lab was among the first to make it work. Engelbart’s October 1962 report for the US Air Force Office of Scientific Research, “Augmenting Human Intellect: A Conceptual Framework,” set out the goal: increasing a person’s ability to approach a complex problem.7

Engelbart had joined the research institute SRI in 1957. From 1963 he had funding for his own lab, which became the Augmentation Research Center.15 Its oN-Line System, NLS, let users work directly with the computer on email, text editing, programming, debugging and hypertext journaling. The Computer History Museum’s guide to the lab’s records, prepared by Elizabeth Feinler, says Engelbart and his group developed the mouse, hypermedia and multiple windows, among other tools.15 The Doug Engelbart Institute says the team used NLS for almost every part of its own work.16

On December 9, 1968, Engelbart and his team showed NLS at the Fall Joint Computer Conference in San Francisco, connected by microwave link to their computer at SRI in Menlo Park.1617 The IEEE milestone for the event, often called the “Mother of All Demos,” lists collaborative online editing, hypertext, video conferencing, word processing, spell checking, revision control and the mouse among the technologies shown.17 The plaque names Brooks Hall as the venue, but a note on the IEEE page says it should read the San Francisco Civic Auditorium.17 About 2,000 people watched and gave the team a standing ovation, according to the Engelbart Institute.16

Berners-Lee later wrote that Engelbart’s work came closest to the web’s design and that he was amazed when he first saw it. Engelbart had used the # sign as a delimiter for a location within a document, as the web does. Berners-Lee says he picked the sign by analogy with an apartment number and learned only later that NLS and an IBM hypertext project had also used it.8

A second early system came from Brown University. From autumn 1967, Andries van Dam and a group of undergraduates built the Hypertext Editing System (HES), with many contributions from Nelson, who later ended the partnership. It ran on an IBM 2250 graphics terminal attached to Brown’s IBM 360 mainframe, which had 512 kilobytes of main memory.18 IBM found a customer for it in NASA, which used HES for Apollo mission documentation.18 Its successor, FRESS, added two-way links and documents with no size limit.18

Hypertext on personal computers: HyperCard and the 1980s

Berners-Lee wrote his own small hypertext program before the web. During a consulting stint at CERN from June to December 1980 he wrote ENQUIRE, short for “Enquire-Within-Upon-Everything,” which linked arbitrary nodes.6 His 1989 proposal says it stored snippets of information and linked related pieces in any way, and that it was similar to a program Apple had produced more recently: HyperCard.1

HyperCard was developed by Bill Atkinson, who announced it on August 11, 1987.1920 The Computer History Museum holds Atkinson’s videotaped overview of it for Apple’s advance briefings, dated July 24, 1987.21

Apple’s own September 1987 description explained the model. Text and graphics lived on cards gathered into stacks, and buttons linked any card to any other card. Buttons could also dial a phone, look something up or print a report. HyperCard was included with every new Macintosh, needed at least 1 MB of memory and came with a scripting language.22 That language, HyperTalk, was created in 1986 and 1987 by Dan Winkler and used English-like commands.20

Thousands of stacks were made with it.19 The game Myst was first released as a HyperCard stack.20 Apple officially discontinued HyperCard in 2004, and in 2017 the Internet Archive made many old stacks usable in the browser by emulating early Macintosh computers.19 The Archive’s account says HyperCard slowly gave way to the growing World Wide Web after about ten years of popularity.19

Hypertext was also becoming a research field of its own. The ACM held Hypertext ’87 in Chapel Hill, North Carolina, on November 13 to 15, 1987. Its papers included Frank Halasz’s “Reflections on NoteCards: Seven Issues for the Next Generation of Hypermedia Systems.”23 Berners-Lee’s 1989 proposal mentions the Hypertext ’87 and ’88 conferences and a standards workshop hosted by the US National Institute of Standards and Technology.1

Berners-Lee’s 1989 proposal: hypertext across a network

In March 1989 Berners-Lee circulated “Information Management: A Proposal” at CERN, and it was sent around again in May 1990.6 It argued that a web of notes with links between them suited CERN, where people and ideas were constantly changing, better than a fixed hierarchy, and it offered its answer under the heading “A solution: Hypertext.”1

What set it apart from HyperCard and most earlier systems was the network. The proposal noted that few products took Nelson’s idea of a wide “docuverse” literally by allowing links between nodes in different databases, and that doing so would need some standardization.1 It also required that existing systems be linked together without any central control or coordination.1

On November 12, 1990, Berners-Lee and Robert Cailliau put forward a formal plan, “WorldWideWeb: Proposal for a HyperText Project.” It described hypertext as a way to reach many kinds of information as a web of nodes that a user can browse at will.3 Each link would be given as an ASCII string from which the browser could work out how to reach the linked document.3 The URL entry covers how those addresses developed.

That choice meant giving up two-way links. Berners-Lee wanted links to other people’s documents to be as easy to make as links within his own, and he later explained the consequence: “Which meant that links had to be one way.”8 Dropping a central link database let the system scale, at the price of dangling links that point to things that are gone.8 Nelson saw it differently. He has written that the web was what Xanadu was trying to prevent, and that it rests on fragile, ever-breaking one-way links with no support for versions.10

In August 1991 Berners-Lee posted the files to the alt.hypertext newsgroup, and in December 1991 the project showed a poster and demonstration at the Hypertext ’91 conference in San Antonio, Texas.6 The story of that period is told in the entry on the first website.

The link itself was written as an anchor tag, <A>. On the project’s 1992 list of HTML tags, an HREF attribute made the anchor text clickable and gave the address to open, and a NAME attribute let an anchor be the target of a link.2

Hypertext on today’s web

The anchor element is still how a web page makes a link. The HTML standard maintained by the WHATWG, updated on October 22, 2026, treats a link as a connection between two resources, one of which is the current page, created by the a, area, form and link elements.24

The standard sorts links into three kinds. Links to external resources point to files used to add to the current page. Hyperlinks are shown to the reader to follow. Internal resource links point to resources within the same page.24 Annotations called link types can change what a link means. The nofollow type, for example, says the page’s author does not endorse the page linked to, and noreferrer stops the browser from telling the other site where the visitor came from.24

The one-way links Nelson criticized also break. In a May 2024 study of a random sample of just under 1 million web pages from the Common Crawl archive, the Pew Research Center found that 38% of pages that existed in 2013 could no longer be reached a decade later. It also found at least one broken link on 23% of the news pages and 21% of the government pages it examined.25

Timeline

  1. Pew finds 38% of web pages from 2013 can no longer be reached25
  2. The Internet Archive runs HyperCard stacks in the browser for HyperCard's 30th anniversary19
  3. A Xanadu document viewer is shown at a Chapman University event14
  4. Apple officially discontinues HyperCard19
  5. Udanax Green and Udanax Gold, two Xanadu systems, are released as open source13
  6. The World Wide Web is demonstrated at the Hypertext '91 conference in San Antonio6
  7. Berners-Lee and Cailliau submit "WorldWideWeb: Proposal for a HyperText Project"3
  8. Berners-Lee proposes a networked hypertext system at CERN1
  9. The ACM's Hypertext '87 conference opens in Chapel Hill, North Carolina23
  10. Bill Atkinson introduces HyperCard for the Macintosh19
  11. Berners-Lee writes ENQUIRE during a consulting stint at CERN6
  12. Engelbart's team demonstrates NLS, including hypertext and the mouse, in San Francisco17
  13. Work begins at Brown University on the Hypertext Editing System18
  14. Ted Nelson introduces the word "hypertext" in a paper at the ACM national conference11
  15. Vannevar Bush describes the memex in "As We May Think"5

What it means for your website

Every website is hypertext, whether it is built on a hosted builder or on WordPress. The links between your pages, and from other sites to yours, decide how visitors and search engines find what you publish.

  • Use real links. Google says it can generally crawl a link only if it is an HTML a element with an href attribute. A link that works only through a script, or a span with an href, may not be followed.26 Builders create proper links for you, but check any custom menus or buttons a developer added.
  • Write link text that says where it goes. Google asks for anchor text that is descriptive, reasonably concise and relevant, and warns against “click here.”26 Under WCAG 2.4.4, a Level A requirement, the purpose of each link should be clear from its text alone or from its text and surrounding context.27 Our glossary entry on web accessibility covers the rest of that standard.
  • Link every important page from another page. Google recommends that every page you care about have a link from at least one other page on your site.26 Our guide to websites and SEO shows how much control each builder gives you over menus and internal links.
  • Check for broken links. Pew found that 38% of pages that existed in 2013 were no longer accessible in 2023.25 Check your outbound links once or twice a year, and when you rename or move a page, redirect the old address to the new one. A website migration is when this matters most.
  • Label paid and user-added links. Google asks for rel=”sponsored” or rel=”nofollow” on paid links and rel=”ugc” or rel=”nofollow” on links that users post, such as in comments.26

For how pages, links, servers and domains fit together, see our guide to how a website works.

See also

References

  1. Information Management: A Proposal. Tim Berners-Lee, CERN (HTML copy hosted by W3C). 1989-03. accessed October 22, 2026.
  2. Tags used in HTML (WWW project pages as of November 3, 1992). CERN WWW project, archived by W3C. 1992-11-03. accessed October 22, 2026.
  3. WorldWideWeb: Proposal for a HyperText Project. Tim Berners-Lee and Robert Cailliau, CERN (hosted by W3C). 1990-11-12. accessed October 22, 2026.
  4. RFC 9110: HTTP Semantics. Internet Engineering Task Force, RFC Editor. 2022-06. accessed October 22, 2026.
  5. As We May Think (scan of The Atlantic Monthly, July 1945). Vannevar Bush, The Atlantic Monthly, via Internet Archive (Ted Nelson Archive). 1945-07. accessed October 22, 2026.
  6. A Little History of the World Wide Web (created by Robert Cailliau, maintained by Dan Connolly). W3C. accessed October 22, 2026.
  7. Augmenting Human Intellect: A Conceptual Framework (SRI Summary Report AFOSR-3223). Douglas C. Engelbart, SRI, via Doug Engelbart Institute. 1962-10. accessed October 22, 2026.
  8. Frequently asked questions by the press. Tim Berners-Lee, W3C. accessed October 22, 2026.
  9. Project Xanadu. Project Xanadu, xanadu.com. accessed October 22, 2026.
  10. Xanalogical Structure, Needed Now More than Ever. Theodor Holm Nelson, Project Xanadu and Keio University. 2000-05-23. accessed October 22, 2026.
  11. A File Structure for the Complex, the Changing and the Indeterminate (scan of the 1965 ACM National Conference paper). Theodor H. Nelson, ACM 20th National Conference, via Internet Archive. 1965-08-24. accessed October 22, 2026.
  12. "The Hypertext": first publication using the word. Theodor H. Nelson, Ted Nelson Archive, Internet Archive. 1965. accessed October 22, 2026.
  13. Udanax. Udanax.com and Project Xanadu. 1999-08-23. accessed October 22, 2026.
  14. After 54 years as vaporware, the web's oldest ancestor is released. Terrence O'Brien, Engadget. 2014-06-06. accessed October 22, 2026.
  15. Guide to the SRI ARC/NIC records (prepared by Elizabeth J. Feinler). Computer History Museum, via Online Archive of California. 2011. accessed October 22, 2026.
  16. The 1968 Demo. Doug Engelbart Institute. accessed October 22, 2026.
  17. Milestones: Public Demonstration of Online Systems and Personal Computing, 1968. IEEE, Engineering and Technology History Wiki. 2017-03-19. accessed October 22, 2026.
  18. A Half-Century of Hypertext. Brown University Department of Computer Science. 2019-05-23. accessed October 22, 2026.
  19. HyperCard On The Archive (Celebrating 30 Years of HyperCard). Jason Scott, Internet Archive Blogs. 2017-08-11. accessed October 22, 2026.
  20. Hypercard Overview. Noel S. Jackling, Museums Victoria Collections. 2007. accessed October 22, 2026.
  21. Hypercard Demonstration, from the Apple Advance Briefings Video Collection (catalog 102639150). Computer History Museum. 1987-07-24. accessed October 22, 2026.
  22. More info: Hypercard (AppleLink, September 1987). Apple Computer, via Known Users archive, savetz.com. 1987-09. accessed October 22, 2026.
  23. Hypertext'87 Proceedings, November 13-15, 1987, Chapel Hill, North Carolina. dblp computer science bibliography (VLDB mirror). accessed October 22, 2026.
  24. HTML Living Standard: Links. WHATWG. 2026-10-22. accessed October 22, 2026.
  25. When Online Content Disappears. Pew Research Center. 2024-05-17. accessed October 22, 2026.
  26. Link best practices for Google. Google Search Central. 2025-12-10. accessed October 22, 2026.
  27. Understanding Success Criterion 2.4.4: Link Purpose (In Context). W3C Web Accessibility Initiative. accessed October 22, 2026.

Web Building Encyclopedia. First published October 22, 2026. Last reviewed October 22, 2026. Every fact on this page is linked to a source we opened. Spotted an error? Write to [email protected].