THE LOST TAGS OF HTML

  1. Nature and purpose of this document
  2. Compliance with the various SGML or XML models of HTML
  3. Sources and Structure of Findings
  4. Technical history of HTML - an Introduction
  5. "HTML 0.a" A Prototype HTML, The Very Beginning
  6. "HTML 0.c" The Earliest Documented HTML
  7. "HTML 0.d" Making the NeXT Editor SGML Compliant
  8. "HTML 0.h" The Oldest Surviving Usable DTD
  9. "HTML 1.k" A First Draft Published for HTML 1
  10. "HTML 1.m" A More Advanced HTML 1 Version
  11. A Brief note on HTML+
  12. HTML 2.0: The Oldest Recognized W3C Version
  13. HTML 3.2: A Much Richer Standard, and a Road Not Taken
  14. HTML 4.01: A Mature and Stable Markup Language
  15. XHTML 1.0: Changing the Basis from SGML to XML
  16. XHTML 1.1: Leaving the Past Behind
  17. A Brief note on HTML 5.0 and XHTML 2.0

"It often happens that the mind of a person who is learning anew science, has to pass through all the phases which the science itself has exhibitedin its historical evolution." - Stanislao Cannizzaro, Italian chemist,1826 - 1910

I think the most truly seminal moment in the history of the internet andthe World Wide Web occurred on that fateful day in 1990 when the decision-makersat CERN Institute in Switzerland accepted a proposal written by TimBerners-Lee regarding a new networking scheme and commissioned him to beginworking on this project, and provided him with a black, cube-shaped computerwith a NeXT editor and other related software on it. Or perhaps it was thatday not long after that Mr. Berners-Lee, using that computer, constructed thefirst working http server and web page featuring an inter-document hyperlink.Though hyperlinks had been mathematically speculated upon for decades, this wasthe first working instance of one capable of linking assets even across acomputer network. Per his proposal, his concern was to provide a way to indexand connect a large and growing number of academic papers being authored bythe various scientists at the CERN Institute, and other scientists around theworld with whom they were in communication through the small but growinginternationally-spread computer network.

Anyone familiar with peer reviewed academic papers knows that one commonmotif within virtually all of them is the footnote, providing a reference tosome other previous academic paper, usually by someone else, to provide somequote or basis for the subject matter being discussed at that point. Inconventional research as it was known back then, if a reader of such a paperwanted to know about something being quoted or referred to, he would look upits title in the footnote, and then go to a library and have the referenceclerk there track down the cited article. This could take hours (if thelibrary has on hand the desired article), or days, or even weeks (if they don't).Already, things were improving in that computers were already being connectedup in a small but rapidly expanding network, and that most current academicpapers of significance were being made available over that network. But eventhere it could take hours to find what you are looking for, and if a paper islarge, to find the desired quote or reference within that paper.

In that day the point and click, mouse oriented, user interface was stillvery much in its infancy, but it was a known idea. Several point-and-clickGraphic User Interfaces were already being marketed. "What if," one canjust see Mr. Berners-Lee thinking to himself, "one could point and clickon a footnoted reference, click on that, and BAM! up comes the particularreferenced article, and even on the correct page within the article wherethe relevant topic is discussed or the quote comes from?" After all, atleast the author of such an academic article would have looked up his ownreferences and would know where to find them, so once he knows where theyare to be found on the network, that information could be saved as (for example)some host_and_file_name field and a computer program("user agent," or "browser") would be able to use thatinformation to fetch the desired article over the network with computerizedspeed and efficiency.

And it would not require all that much time, despite the slowness (as wewould see it today) of the network in 1990. Bear in mind that in 1990, datatransfer rates were measured in "baud." A common baud rate found inthat era was 9600. Sometimes one could get data transferred at 19200 baud, butthat was still rare and error-prone, and some nodes on the network were limitedto far lower baud rates. On the other hand, an academic paper did not have alot of additional garbage (graphics, applets, advertising banners, and justoverall lengthy and cumbersome formatting tags) that so dominates virtually allweb pages today. They would be formatted as raw ASCII text files, and look verymuch like this online copy of RFC 1111. So even at 9600 baud, alarge article like this could still download within a matter of seconds, farfaster and easier than going to the library, or even having to track it down onthe network, and then use ftp: or telnet: to downloadit or look at it, as was the common practice in 1990 when using the computernetwork for academic or scientific research.

As there was little to no software for "Web programming," much of thisinitial HTML research and experimentation was done using the NeXT editor or elsesome simple ASCII text editors. HTML was meant to be easy to write, simple in whatit does and how it does it. The practical upshot of that background is theinteresting result that HTML is the easiest "computer language" to writein, by far. It is so easy that just any Joe Sixpack can pick up a basic HTMLHow-To book, and with a computer, internet connection, and simple text editor, hecan be writing and posting presentable HTML web pages within half an hour.

Unfortunately, as HTML advanced, this simple charm of the easy-to-master HTMLlanguage gets gradually lost as HTML not only acquires features, but becomesmore and more dependent upon them, and also as basic features that made it soeasy are done away with, being replaced with more advanced and versatilefeatures. The Joe Sixpack who can be coding up presentable HTML 2.0 so quicklywould be at quite a loss trying to do the same with XHTML 2.0 in anything like thesame short period, even once there should actually come to be such a thing as anofficial published XHTML 2.0. The more HTML is made more compatible and similarto the other file formats, such as Microsoft Word documents (.doc), RichText Format (.rtf), or Portable Document Format (.pdf), theless human readable it becomes, and the more the document writer becomesdependent upon special editors to do all of the low level stuff. Correspondingly,the files themselves also become less compact as more and more sophisticated andobscure (and lengthy) non-text data comes to predominate, to say nothing of theway these editors generate inefficient and wordy/lengthy code to begin with.

There is also the question of whether these HTML features should be called"elements" or "tags." Originally, the custom was to call them"tags" but as HTML as a language got tightened up and formally documentedthe alternate expression "elements" seems to have displaced it. For awhile there was a tug of war between these two terms, but in the end, theconsensus seems to be that, one uses the word "element" to refer to theitem in general, as a Platonic or philosophical ideal, or (per the document objectmodule and style sheet functionality) the whole package of an opening tag, theclosing tag (whether present or implied by context), and all the material betweenthem, and the word "tag" to refer to an occurrence of an element, or moreparticularly its opening and/or closing tags, within an HTML file. So, forexample, one would say, "The <A> element serves to providea link to another document, or another portion of the same document," or else"This <A> element links the phrase XYZ to thedocument ABC," and "Here is the problem; you forgot to put aclosing </A> tag here in your file." In this treatise, Iwill tend to lean towards the older and more colloquial usage of "tag" inmost places. Fortunately, the term for the other relevant HTML feature which Ialso discuss, the "attributes," has never gone through such name changecontusions.

Here Lies <TAG> R. I. P.

The approach of this online HTML treatise is to focus mostly on those particularelements ("tags") which are gone, or at least going away, and to track indetail the comings and goings of the various elements and attributes across theversions of HTML. A special emphasis is given here to those many versions of HTMLthat precede HTML 2.0, a breakdown of the many discrete versions of HTML, which issomething that to my knowlege has not yet been attempted. Most modern HTMLwritings focus either on those elements with long-standing value (such as<A>, <TITLE>, <UL>, and<P>) as necessarily continue from the very earliest HTML to the(as yet) latest version, and on the brilliant and powerful new features whichare being added. Little to no attention however gets paid to thoseother elements ("tags") which are falling by the wayside, andwhich altogether don't exist in XHTML 1.1 and later versions of XHTML, or else aretolerated in HTML5 only as backward compatibility for old files, and even less tothose already declared obsolete in HTML 4.0. (One notable exception is HTML& XHTML - The Definitive Guide, by Chuck Musciano & Bill Kennedy, O'ReillyMedia, Inc. which retains some of this early material.)

For myself, I find that the unique perspective of studying those tags andattributes which are depreciated, going away, or completely gone should prove tobe an interesting approach to exploring the history and nature of HTML, andoccasionally a few of some of the related technologies. Even so, I don'tcompletely restrict my considerations to the lost tags, but also take thisopportunity to explore some others of the more arcane and obscure features of HTML,the kind of stuff that receives little to no coverage, original material notavailable elsewhere. But this treatise is not meant merely to serve as a nostalgiatrip down memory lane (though it may serve as that too). Rather, it is a chance todiscuss the nature of technological change and progress, to see some of theacademically clever ideas that fail to take hold in the marketplace, and converselythe ideas hastily implemented by the industry but with little or no standing withthe academic community due to their misuse of what HTML was really meant to beabout, to discuss backward-forward compatibility issues, and to understand some ofthe inner workings of HTML from a layman's perspective. Furthermore, support forthese depreciated and obsolete tags and attributes continues to be, in many of themore contemporary browsers and user agents, done with grave inconsistencies. A newgeneration of browser writers has arisen who have no real understanding of whatthese underdocumented tags and attributes were, what they did, or how they weremeant to be used.

This is not meant to be the sort of "directed research" guide thatprovides direct and easy access to how to code up the latest new HTML widget,but rather intended to serve as a kind of "non-directed" research toinspire curiosity, interest, and perhaps one may serendipitously come acrosssome clever HTML idea. In short, this may not be what you want, but at leastit should be interesting.

The structure of this work is that it all starts from this main file, whichitself is a simple example of fully compliant XHTML 1.1, transmitted asContent-Type "application/xhtml+xml," and from whichindividual files discuss each of the specific tags, attributes, or topics ofinterest here. An XHTML1.0 Strict version of this file, transmitted asContent-Type "text/html," is located here. The subfiles are renderedin the various versions of HTML or XHTML, in most cases transmitted asContent-Type "text/html," and themselves examples of thetag or attribute or topic in question. These files in turn discuss the history,usage, and recommended implementation of these various tags and attributes, andfinally include further subfiles of their own as examples of upgrades anddowngrades. The Upgrades and Downgrades discuss alternate means of achieving asimilar effect as the tag intends using either more or less advanced features,and also various additional features to be added to the tag in later versions ofHTML or even XHTML. Optionally, there may at times also be a discussion of somefew selected "proprietary extensions" relating to the tag or version ofHTML being discussed, but this consideration is by no means intended to beexhaustive.

Though the reasons for deleting these lost tags and attributes from thenewer versions of HTML and XHTML are explained herein, that is not to say thatsuch omissions are justified in browser implementations. The Joe Sixpack HTMLprogrammers of the world are not going away anytime soon, and even now areapplying pressure to the Web Consortium to please develop HTML and not merelyfocus on XHTML, since HTML programming is so much easier to do. These samefolks also so often use older and simpler versions of HTML and its explanatorydocuments that it is best that the deleted tags and attributes continue to beimplemented in browsers as many of them indeed continue to be. And if theyare to continue to be implemented, I discuss here the optimal manner in whichthey ought to be implemented that best captures the spirit and intention ofthem as originally introduced.

This work can only exist online since it provides actual working examplesof these lost tags and attributes. This serves to demonstrate what they werelike, to enable one to test how their own browser handles these things, to showthe full and proper context in which they function, and to do all this withfiles that conform to the various HTML standards. This is a "work inprogress" as is all HTML, so subjects not treated of herein as yet will beadded as time progresses.

Compliance with SGML or XML models of HTML

With the two exceptions of a) examples that feature the various proprietaryextensions, or b) some very early tags and structures which disappeared beforeHTML 2, I have endeavored to see to it that all examples herein comply fullywith the World Wide Web Consortium's standards. Each file can be validated byusing a link at the bottom (or else a special link to its validation in itssubfile above), and which not only validates the file but also shows its sourcetext exactly as validated. The following standards arevariously used, depending on the file:

In the case of those examples that feature proprietary extensions, I haveseen to it that with the one exception of the particular proprietary tag,attribute, attribute value, or nesting of the tags that cannot comply with anyof these standards, the file is nevertheless otherwise in compliance with oneor more of these standards. In the case of certain tags and attributes thatvanished prior to HTML 2.0, I have written it to the standard that introducedthe tag or attribute, or else in a "standard" which appears to reflectthe nature of HTML at a given time, as captured in DTDs that I have written.

Many HTML guides out there showcase examples of at least some of thesimpler and commoner tags which display reasonably on most browsers, buthere we try to showcase the usage of somewhat more complicated tags to useand also try to take care to see to it that it complies with whichever HTML(or XHTML) standard it needs to comply with, or specifically note where itdoes not or cannot comply.

One convention I use here is that every file which is compliant withsome version of HTML will begin with the <!DOCTYPE>declaration appropriate to the version of HTML it complies with. Thosefiles with proprietary details which fail to conform will lack thisdeclaration, but the link I provide to validate (and fail) the file willalso automatically specify the standard I otherwise wrote it to. Though Icall many of these files "Working Examples," in some cases there isno discernable effect in the display or operation of the page. In suchcases, the only thing that "works" might be the ability to validatethe file containing the rare and otherwise almost never seen element. The<!DOCTYPE> declaration itself however is not a part ofHTML but rather of SGML, the basic language behind HTML. Each version ofHTML (except some of the earliest) is described in a special text file calleda DTD (Document Type Definition). A DTD is written in a specialmachine-and-human-readable SGML format which specifies the details of theversion of HTML being validated. A couple small parts of the DTD for HTML2.0 would look like this:

<!-- Modified for use in HTML$Id: ISOlat1.sgml,v 1.2 1994/11/30 23:45:12 connolly Exp $ --><!ENTITY AElig CDATA "&#198;" -- capital AE diphthong (ligature) --><!ENTITY Aacute CDATA "&#193;" -- capital A, acute accent --><!ENTITY Acirc CDATA "&#194;" -- capital A, circumflex accent --><!ENTITY % linkType "NAMES"><!ENTITY % linkExtraAttributes "REL %linkType #IMPLIED REV %linkType #IMPLIED URN CDATA #IMPLIED TITLE CDATA #IMPLIED METHODS NAMES #IMPLIED "><!ELEMENT LINK - O EMPTY><!ATTLIST LINK HREF CDATA #REQUIRED %linkExtraAttributes; %SDAPREF; "Linked to : #AttVal (TITLE) (URN) (HREF)>" ><!-- <LINK> Link from this document --><!-- <LINK href="..."> Address of link destination --><!-- <LINK URN="..."> Lasting name of destination --><!-- <LINK REL=...> Relationship to destination --><!-- <LINK REV=...> Relationship of destination to this --><!-- <LINK TITLE="..."> Title of destination (advisory) --><!-- <LINK METHODS="..."> Operations allowed (advisory) -->

One can see from this how the <!DOCTYPE>declaration found in an HTML file is actually an SGML command, one which tellsthe validator which particular version of HTML is being scanned. Illustratedabove are the SGML DTD file entries to specify that the expressions &AElig;,&Aacute; and &Acirc; will result in Æ, Á, and respectively, and that the <LINK> tag takes the attributesof HREF, URN, REL, REV,TITLE, and METHODS. It also specifies that theHREF attribute is required (must be present in any occurrence of the<LINK> tag). So, for example, the HTML validator would lookat this file to know the language details of the particular version of HTMLyou are validating to, and then scan your file. If it found in an HTML file a<LINK> tag with some other attribute, not on that list, orelse a <LINK> tag which is lacking the HREFattribute, it would flag that as an error. So you can see that just as<!DOCTYPE> is used to specify the type of file you have (abit of SGML in an otherwise HTML file), <!ENTITY> in theSGML DTD file can specify a special character display command (as it does here)or a DTD-internal string definition (as also seen with the linkTypedeclaration above, <!ELEMENT> specifies a tag type, and<!ATTLIST> specifies the list of attributes the particulartag (or "element") takes. Such entities as "&lt;"are themselves actually SGML constructs. Another bit of SGML commonly seen in HTML filesis the comment (<!-- The contents of this wouldn't display. -->),and occasionally other SGML structures may turn up as well. Many of such othersorts of SGML-distinct items may not display well in most user agents, as forexample demonstrated here.

There does exist a workaround by which some of the more primitive types offiles can be validated, and that is by using special document definition files todescribe these more ancient versions of HTML. I have made some of these availableusing the <!DOCTYPE> declarators mentioned below. Note thatwhen the W3C validating engine is used with any of the older DTDs it will issue awarning to the effect that the Document Type is not in the validator's catalog.I have a small note here regarding details of theDTDs that I have prepared or adapted for these unofficial versions of HTML, sothat they can be experimented with in detail.

Sources and Structure of Findings

There are three basic sources of information as to the most earlyversions of HTML. The first and most difficult to use would be the vaststore of actual files from the early periods as archived on the WebConsortium's own historic archive files. For this one can only scan alarge number of files (or as many as can be found to date from a periodof time) to glean the actual nature of the particular variant of HTML inuse during said period. The second source would be those few attempts todocument HTML as it stood at this or that particular point in time, orelse to suggest or propose various extensions or refinements to HTML.The third and most reliable source would be the actual DTD drafts by whichvarious early versions of HTML are precisely defined. There was not at thattime any attempt to track or identify any particular "versions" ofHTML, and there is no official "HTML 1," but the HTML draft publishedin mid-1993 is often taken as being a kind of "HTML 1." I draw theline between HTML 0 and 1 on the basis that only with HTML 1 were the DTD'sbeing specifically associated with, and intended to define most formally, whatwas or was not in the current working draft of HTML. In HTML 0, though thereexisted some DTD's in the latter part of the HTML 0 period, they existedmostly as research exercises, running only in parallel to the general run ofHTML as used and intended. I have assigned continually ascending lettersfor each of the versions, leaving room for some few additional versions toturn up, with the additional provisions that a through e wouldbe for those which have no DTD but are known through actual files or generaldescriptive documentation only, and that f and beyond are for theDTD-based versions of HTML, k and beyond for HTML version 1, andthat p and beyond for HTML version 1+ ("HTML+").

As a result, I have identified a number of "versions" of HTML,many of which I document here in some detail, and which existed prior toHTML 2. And I believe there were probably more other versions as well asthese, which may turn up someday. Since there was no strict versioningmechanism prior to HTML 2, I have arbitrarily assigned the followingdesignations to the various "HTML's" that preceded HTML 2 that Ihave identified (those in bold are described in detail below; the twosurviving versions of HTML+ and the Draft of HTML 3.0 are also given somedescription here):

  1. HTML 0.a - That most primitive prototype of HTML used by TimBerners-Lee and a few of his closest associates in "The Project"from the very beginning in October 1990 to mid-January 1991.
  2. HTML 0.b - Unassigned, but it could be seen as a kind of variant ofHTML 0.a seen towards the very end of its period, in which <P>briefly goes from being a separator tag to a container tag, but after whichit goes back to being a separator tag until the second release of HTML 1,and in which <OL> seems to be at least depreciated.
  3. HTML 0.c - That form of HTML that spans a period from January 23, 1991through November 23, 1992, and for which documentation was written in itsown version of HTML towards the end of that period.
  4. HTML 0.d - That form of HTML that spans a period from November 26, 1992through May 24, 1993 (or thereabouts), most specifically characterized bythe use of the <HEADER> tag to set apart its head.
  5. HTML 0.e - That form of HTML which Dan Connolly described in lateNovember 1992, and wrote some few files in, which frequently contains the<TYPEWRITER> tag and/or a <NEXTID>attribute called ID.
  6. HTML 0.f - The HTML of the oldest surviving DTD from August, 1992.
  7. HTML 0.g - That DTD (now lost) which was pointed to by the November1992 documentation of HTML; I conjecture that this DTD might have calledout the <TYPEWRITER> tag before going to<PRE> as seen in the next version.
  8. HTML 0.h - The HTML of the oldest usable DTD from January, 1993.
  9. HTML 0.i and 0.j - unassigned, room for possible expansion.
  10. HTML 1.k - The first released draft of HTML ("HTML 1") inJune, 1993
  11. HTML 1.l - unassigned, might apply to a variant of HTML 1.k that fixesdifferences from as described in accompanying text, for example addingALT and ALIGN attributes to the<IMG> tag, or possibly a first introduction of the<META> tag.
  12. HTML 1.m - The second draft of "HTML 1" in May, 1994.
  13. HTML 1.n and 1.o - unassigned, room for possible expansion.
  14. HTML 1.p - Lost early draft of HTML+ from late-mid 1993.
  15. HTML 1.q - Oldest surviving draft of HTML+ in November, 1993.
  16. HTML 1.r - unassigned, room for possible intermediate version ofHTML+.
  17. HTML 1.s - Latest surviving draft of HTML+ in April, 1994.
  18. HTML 1.t through 1.z - unassigned, room for possible expansion.

After these comes the official (and sometimes not-so-official) "versions"of HTML, and again those in bold are treated of in detail below:

  1. HTML 2
  2. HTML 3.0
  3. HTML 3.2
  4. HTML 4.0
  5. HTML 4.01
  6. XHTML 1.0
  7. XHTML 1.1
  8. HTML 5.0
  9. XHTML 2.0

In these quick listings of the tags (elements) and attributes of each versionof HTML I am using here, I have utilized the following color scheme to representthe various categories of the state of each tag and of each attribute, so as tomake comparison between the various versions easier at a glance. These are thecolors used:

The algorithm used in calculating the color is thus: If a tag or attribute isbeing newly introduced with the version being given here (I disregard if anyintermediory version, such as HTML+, HTML 3.0, or HTML 4.0, may have introduced it), itwill be either Green or Chartreuse,depending on the next question. If a tag or attribute is any of a) listed or describedin the W3C document version as being "depreciated" (or even "obsolete"as seen in some earlier texts), but still present in the language, or b) not presentin the "strict" form of the language, or c) omitted from the next version ofthe language, or d) confined to a "Legacy Module" (in XML versions), or anycombination of the above, it will be Red or Chartreuse, depending on the previous question. If a tag orattribute, present in any previous version of HTML, is missing from the presentversion, it is Faded Blue. If none of the aboveconditions apply, it is Black.

A Technical history of HTML

Described below are the versions of HTML which I have identified and discussed indetail. The earliest three versions that I have identified have been based upon thearchive store of ancient HTML files found in the historical archives of the WorldWide Web Consortium (W3C) website, looking in detail at the nature of the HTMLcontained in the files and tracing the history sequence by using the date modifiedinformation contained in the HTTP header information for that file, as archived bythe W3C. These ancient files contain a vast repository of the earliest efforts ofTim Berners-Lee, Dan Connolly, and others. By observing these files, I haveattempted to construct what I call "simulated" HTML DTDs for each of them.As it turns out, there are two problems with the earliest two of the versionsthat render it impossible to validate them by SGML. For reference and ease ofdistinction I have given these first three attempted versions the names of HTML0.a,HTML0.c, and HTML0.d (only the DTD for this last one can work).

In addition, I have found six DTDs older than the HTML2 DTD. The oldestsurviving DTD I have found is from August 20, 1992 and it was written by Dan Connolly.This early DTD is far too incomplete to regard as a working version of HTML, thoughit did point to some future directions HTML would soon take. For reference I call itHTML0.f. The next DTD to survive was also written by Dan Connolly, on January 20,1993. For reference I call it HTML0.h. Then next came the first released draft foran official HTML 1 DTD published (on the web) on July 01, 1993 (which I call HTML1.k)and a second release of HTML 1 which occurred on May 18, 1994 (which I callHTML1.m). Finally, there are two HTML+ DTDs, one from 1993 (but last modified inApril 1994, and the other from about April 1994 (but last modified in 2000), butthese DTDs, written by Dave Raggett, follow a very different approach (more akin tothat of HTML 3.0 than HTML 2) and cannot be logically fitted into a sequentialposition, and introduce much that is not seen in any other version of HTML.

There are reports of DTDs being in work by various people as early as May 1992,and Dan Connolly's August DTD mentions a first pass at making a DTD dated 15 July,and the November 1992 descriptions and discussions of HTML pointed to a DTD whichwas current at that time (but is now lost). The pointer is still active, butpoints to a file last modified in late 1995 and contains what is essentially apenultimate draft of HTML 2. I doubt however that it ever exactly matched the descriptionof HTML given at that time, and this seems to be a continuous problem in thedocumentation of HTML. Usually due to proofreading or typographical errors in theDTD the description of the version of HTML given in the official W3C documentscontains some few differences from that actually valid per the DTD. Except for thefirst three versions, for which no official DTDs exist, in the tag and attribute listsgiven here, it is the DTD and not the narrative documentation which has been followedas the source of what actual tags and attributes actually exist in the level of HTML.Such differences will be noted as they arise.

As the HTML languages get more advanced, the number of tags andattributes expands tremendously, so what I propose to do here is list them in"clusters." Many tags belong in some specific category of tag, orelse are used only with a small group of related tags. As we transition tomore and more advanced versions of HTML we will add clusters and also add tothe existing clusters already introduced in the older versions.

"HTML 0.a" A Prototype HTML, The Very Beginning

The very oldest known surviving HTML file was last modified on Tue, 13 Nov 199015:17:00 GMT. Its complete text reads exactly as follows:

<title>Hypertext Links</title><h1>Links and Anchors</h1>A link is the connection between one piece of<a href=WhatIs_html>hypertext</a> and another.

This can be found here.Following its link brings one to many other pre-HTML 2 files, many of which, datingfrom before 1993, can also be found here. These veryoldest HTML files from the first four months or so of HTML all bear a number ofdistinct characteristics that make them detectible as further examples of the veryearliest form that HTML took. This earliest form of HTML was never documented but itcan be seen and gleaned from the following files that were programmed in it and whichall stem from the earliest period for HTML. They can be distinguished by the use oflower case for the tags (similar to XHTML) no <nextid> tag, andoverall very primitive use of only a very small group of the oldest and most originalHTML tags. It also appears to have been hand-entered, since several typographicalerrors have been detected in the HTML tags. Occasionally, such hand-entered HTML wouldsurface again, distinguished by the same characteristics, sometimes even in filesgenerated by the NeXT HTML generator, in what places they have been manually retouched.

It appears that Tim Berners-Lee only wanted to spend a very few days extracting onlythe most basic and rudimentary functions from CERN's SGMLguid language for this smallprototype markup language, so that he could instead devote his time to the far larger andmore complex issues of creating the kind of network and servers and overall infrastructurethat would make his "link" idea work. As such this first crude pass at HTML wasaimed only at providing a basic framework of document formatting/structuring commandswithin which to position his true brainchild, the <a href=Filename_html>link. Obviously he intended from the start that this markup language would soon be expandedupon by both himself and others, and things did indeed unfold just that way. Thefollowing file showcases the look and source of all such files that have been identifiedas having come from this earliest prototype HTML period: Sourceand Contents of the oldest surviving HTML Files. Basically, most of the remainingsuch files are documentation (papers, requests for papers, announcements, trip reports,tutorials) pertaining to the European Conference on Hypertext, 1990 (ECHT90) which tookplace on 27-30 November 1990, and these files were last updated December 1990 and thefirst half of January 1991, with the exception of the last four which, though alsodating from this period, recieved minor updates, one in late 1991 and the other threein mid-1992.

Here is the set of tags and attributes for HTML as it had existed during its firstseveral months, from its beginnings in October 1990 until mid-January 1991 when thenext style appears:

Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
Listing Tags
Dictionary Tags
Inline Formatting Tags

Even this list, very short as it is, may be somewhat artificially expanded with theinclusion of <xmp> and <listing> and the dictionarytags which, though seen in a test file from the period, do not otherwise seem to havebeen used in any other files from this period and which may not have been implemented.Clearly, the inclusion of these otherwise unseen tags within the one test file signaledan intention that these would one day exist (and the file does show them), if not atfirst, then at least eventually. Though the last two of the sample files also includethe dictionary tags, both of them have been modified on August 26, 1992, quite possibly toadd these dictionary tags, and in one of which these tags (<DL>,<DT>, and <DD>) appear in upper case while allother tags in the file are in lower case, showing that they were indeed entered by a laterhand some time after the original creation of the file. The <ADDRESS>tag is seen only in one file modified on June 11, 1992 and not included in the test filenor otherwise mentioned, so I opine that that tag is not native to this early period.

The difficulties with producing an SGML DTD for this initial period are that itdid not need (perhaps did not even take, as there are no counterexamples) any quotationmarks to surround the attribute values. So by SGML standards, any href or name valuecontaining either a "/" or a "#" or a"@" would cause an error with parsing-checking engines. Furthermore,the <p> tag appears to have been variously used, at first being used as aparagraph separator and empty element (much like <BR> would later come tobe, but producing a visible gap to separate the paragraphs), but then towards the end ofthis period as a container, designed to have text "contained" between an opening tagand a closing tag. Some </p> closing tags have been spotted in thesemost ancient HTML files. Starting with the next version of HTML the <P>tag occurs strictly as a separator tag clear until the second draft of HTML 1.

All the tags and attributes from this era, except for the provisional and largelyunimplemented <hpn> tags, have proven to be of lasting value,although one of them would quickly have the distinction of being the first useful tag togo away. <ol> is commonly spotted in this period, but was deletedfor the next version of HTML, and only appears to have been reintroduced with the DTDsby Dan Connolly, and finally got accepted back in by Tim Berners-Lee for the DTD forHTML 1.k. The <ol> tag has been a useful workhorse tag since itsreintroduction, and remains current to this day along with all other tags from thisinitial period. Though the <p> and <li> tagswould continue to modern times, their usage has changed a bit. While in this earliestperiod they served as separator tags (except for <p> which brieflyserved as a container tag) with the next version, and clear into HTML 1 they (alongwith the dictionary tags <dt> and <dd>) began toserve exclusively as separator tags, only to be all changed to container tags for thesecond draft of HTML 1, and remain thus to modern times. At this point, the<title> tag was clearly considered optional, given that so many ofthese early files omit it.

I have constructed a DTD for this file type and it can be used for validating files byaffixing the following declaration to a file and submitting it to the W3C validatingengine:

<!DOCTYPE HTML SYSTEM "https://backiee.wasmer.app/http_www_the-pope_com/html0.a.dtd">
  1. Working XMP and LISTING Tag ElementExamples
  2. Working Tag Elementexamples of HP0, HP1, HP2, andHP3

"HTML 0.c" The Earliest Documented HTML

One of the very oldest files that shows signs of being produced by the NeXT HTMLeditor is thisone dating from 23 January, and this file is the oldest surviving use of the <NEXTID>tag. There is also a filenear the end of January 1991 which seems to be from this era since all tags are in uppercase, but it lacks a <NEXTID> perhaps merely because there are noname links in the file so NEXTID would have been set to zero, the same asnot having a NEXTID value to the file. This oldest surviving description of HTMLwas mostly written in the version of HTML it describes, but it is still helpful tosee a random sampling of the surviving files written in this version of HTML, as shownin Source and Contents of a sampling of the earliestdocumented HTML. Even that description is not totally precise in that it listsone tag which was as yet not in evidence and probably did not exist, namely the<BASE> tag, and one attribute, the TYPE attribute of<A> which also is not seen, anywhere. Here is the set of tags andattributes for HTML as it developed and was used over the course of most of 1991 and1992:

Document Level Tags
Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
User Input Tags
Listing Tags
Dictionary Tags
Inline Formatting Tags

This short list is also artificially lengthened by several placeholder tagslisted there more as room for expansion rather than realized functions. The<ISINDEX>, <DIR>, and <MENU> tagswere described as existing, but appear to have arisen quite late in this period, andno examples from the period of these tags have been found. At this point in time,<NEXTID> did not take an attribute proper, but simply a number itself(e. g. <NEXTID 3>). Conspicuously absent at this point were all documentlevel tags other than <PLAINTEXT>, which is another reason why HTML hadto make those tags (<HTML>, <HEAD>, and<BODY>, and their closing tags) all optional. Though they did not as yetexist in this version, they too were among those proposed and being discussed.

I am sure that this "version" of HTML did not all burst upon the scene fulland complete in January 1991. Some tags, such as <NEXTID>, indeedgo to the very beginning of this period, while others such as <DIR>and <MENU> obviously arrived later on. The dictionary tags(<DL>, <DT>, and <DD>) appear toto have been implemented by the end of January 1991, while the document level tag<PLAINTEXT> and the the block-level raw text tag<LISTING> are reported as having been implemented in at least somebrowsers by as early as February and March of 1991, though no examples of either of these(apart from demonstration files) are found until months later on. Due to the fact that the<ISINDEX> tag was only to be generated by a program runnning on the server,and such executable programs were seldom preserved and doubtless would not even run oncontemporary machines anyway, there is no way to tell what use if any it had during thisperiod, or when it was introduced. During the course of this period, the<TITLE> tag would rapidly transition from being optional to being requiredfor an HTML document. It is not clear that the <HPn> tags were everactually implemented as no example of them have been found apart from the test file fromthe previous period. A direct equivalent of that test file, but stemming from this period,omits the <HPn> example instances, together with an omission of the<OL> element contained in its older form from the previous period.

This phase of HTML represents a clear distancing from SGML and the CERN SGMLguid languageas anything but a potential source of some possible tag names and meanings (e. g.<Hn>). Such a period does seem to precede a certain academic disciplinebeing applied to HTML, almost more of a "Joe Sixpack" stream of consciousness mode ofprogramming, rather than any rigorous academic structure. In conceiving this version of HTML(mostly back in January, 1991), it seems to have been uncertain as to what the basis for anyacademic discipline would be applied to HTML in order to define it as a language. Instead,just any old "good idea" that happened to come along would be readily added to thelanguage as a new and useful feature. Later on, a distinction would gradually form betweenthose structures and elements and attributes coming from the industry versus those comingfrom the academic community, but at this point everything was welcome.

So, when this version was developed, SGML and SGMLguid had pretty much become merely onemore source among many for some "ideas" or "influences" from which to drawsome useful features. The <NEXTID> tag in particular, as seen and also aswritten up in the documentation at that time, merely took a number instead of a formalSGML-type attribute. This is bad SGML and there is no good way to build a DTDto define this version of HTML on account of this one thing alone. (For the "html0.c"I generated with which to validate this kind of file, this <NEXTID>feature was approximated by making the attribute an enumerated list of possible explicitvalues, which only go up to 127, so any value higher than that will not validate. However,no files from this period appear to have ever gotten to so high a value for this to be aproblem.) This is also the phase in which another SGML-hostile tag would be introduced,namely <PLAINTEXT>.

I have constructed a DTD for this file type and it can be used for validating files byaffixing the following declaration to a file and submitting it to the W3C validatingengine:

<!DOCTYPE HTML SYSTEM "https://backiee.wasmer.app/http_www_the-pope_com/html0.c.dtd">
  1. Working NEXTID Tag ElementExample
  2. Working PLAINTEXT Tag ElementExample
  3. Working ISINDEX Tag ElementExample
  4. Working DIR andMENU Tag Element Examples

"HTML 0.d" Making the NeXT Editor SGML Compliant

The W3C historical archives contain no NeXT-HTML-Editor-generated files datingfrom either November 24 or 25 1992. Perhaps Mr. Berners-Lee spent this timeinstalling and shaking the last bugs out of a significantly new version of the NeXTeditor which implemented the first rigorous attempt to comply with SGML as the formalbasis of HTML as a language. All files in the archive dating from November 23 of thatyear and earlier are as described in the above paragraphs here, but files dating fromNovember 26 until at least as late as May 24, 1993 bear the characteristics of thisnext version of the NeXT editor, except where obviously generated earlier and thenhand-edited during that time, or else generated by someone else who had not as yetloaded the new version of the NeXT HTML Editor.

By May 1992 it had been pretty much decided that SGML would provide the academicbasis for this new computing invention called HTML. Nearly all structures in thelanguage could be expressed and defined in an SGML DTD, leaving only a few detailswhich could not be defined in SGML and so which had to be changed. Also, a numberof new ideas were being thought of, and drafts of the DTD easily incorporated manyof these ideas. So important were these ideas that a couple of them were actuallyincorporated into the description of HTML as last modified on November 13, 1992.There was the <BASE> element which would not be implemented oreven defined in a DTD until HTML 1. And there was also the TYPEattribute of <A> which had no specific list of possible values,only some general discussions as to what sorts of possible values might later onappear. It too was only a placeholder within the documented description of HTML.

There is no formal documentation for this next phase of HTML, only a number offiles saved as examples of it in the W3C archives,and a few files that document the Discussions forFuture Directions for HTML. In the course of these discussions going on at thattime, there were vague proposals which would later lead to the introduction of the<LINK> and <PRE> tags, and there was talk ofrestoring the ordered list tag (<OL>). These discussions evenincluded the possibility of a <DATE> tag which might even featurean EXPIRES attribute, and a <KEYWORDS>, tag toassist automated searches, which together gradually matured and finally emerged inHTML 2 as the far more versatile and useful <META> tag. The tagsobserved for this period (or reasonably inferred) are as follows:

Comments
Document Level Tags
Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
User Input Tags
Listing Tags
Dictionary Tags
Inline Formatting Tags

Unlike the previous versions, a DTD can be constructed to validate files of thisinterim standard in accordance with SGML principles. In particular, the<NEXTID> tag recieved its attribute N, making itat last SGML compliant, the <PRE> tag had been introduced, andgot considerably more use in this period than the previous non-SGML tags itreplaced (though I am sure that the non-SGML tags were all still available). Alsoat this time appears the beginning of a formal document structure, by separating theHTML document in to a head (<HEADER>) and body(<BODY>), and confining <TITLE>,<NEXTID>, and <ISINDEX> to the head and therest to the body. Note however the different name for the head tag than wouldappear in all future versions of HTML. I have constructed a DTD for this filetype and it can be used for validating files by affixing the following declarationto a file and submitting it to the W3C validating engine:

<!DOCTYPE HTML SYSTEM "https://backiee.wasmer.app/http_www_the-pope_com/html0.d.dtd">

The only "Lost tag" to be introduced during this phase is the<HEADER> tag, which is exactly replaced with its lasting successor<HEAD> in all future versions. This could almost border on thelevel of a proprietary tag, since it seems to have belonged to the NeXT editoralone, but since it is the earliest, and since so many W3C archive files from thisperiod bear this unusual tag, I have included it in my successive listings. It andthe <HPn> tags are the only tags I list here whose disappearencewas so long ago that they show up on no surviving DTD. These tags therefore havethe distinction of being the very first truly Lost Tags. Though <OL>disappeared earlier, its subsequent return, complete with even the same form fit andfunction, removes it from the category of Lost Tags.

  1. The First Lost Attribute TYPE ofA and LINK
  2. A Look at theExperimental Tags/Elements and Attributes

"HTML 0.h" The Oldest Surviving Usable DTD

By 1993, it was pretty much accepted that SGML would indeed be the basis for thenew hypertext document format language, HTML, and that an SGML DTD would be theofficial lexical definition of HTML documents. Though some several DTDs werewritten in 1992, only a rather incomplete and preliminary one of them has survived.After that, the oldest surviving DTD was writtenby Dan Connolly and dated January 20, 1993. Contained in this DTD are the twoearliest attempts to insert stylistic formatting instructions into the HTML, namelythe WIDTH attribute of <PRE> and a STYLEattribute (never to be seen again in this form) of the <DL> tagwhich would soon morph into the COMPACT attribute of all listing tags,<UL>, <OL>, <DIR>,<MENU>, and <DL>.

Comments
Document Level Tags
Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
User Input Tags
Listing Tags
Dictionary Tags
Inline Data Type Tags
Inline Formatting Tags

This earliest surviving usable DTD seems consistent with what one would expect asthe logical progression from the state of things in November 1992 when the HTML asit had been for nearly two years was documented and also the first SGML-compliantversion of HTML to be generated by NeXT also appeared. Not only does this replace<HEADER> with <HEAD> and return <OL>to the language, but more importantly it adds so many basic inline text typeswhich would prove of lasting value, as a far superior replacement of the old<HPn> tags, such that they are no longer found (if indeed theyever occurred in any DTD). Even so, three of these tags would disappear by thesecond release of an "HTML 1" draft, <KEY>,<DFN>, and <U>, but the latter two would returnin HTML 3. It is in this phase that the WIDTH attribute of<PRE> would be introduced as the first temporarily survivingpresentational feature, though a companion presentational feature STYLEof the <DL> tag would soon mature into the equally temporarilysurviving <COMPACT> attribute of the listing tags, both of whichwere only implemented in some very few early browsers. Drawing from the academicand more theoretical minds, this is the point at which the URN andMETHODS attributes of <A> and <LINK>would be introduced, only to disappear promptly after HTML 2, but also the usefuland surviving TITLE attribute of the same tags. <LINK>is finally introduced here, and in this initial form given simply the exact sameattributes as <A>. By HTML 2, the NAME attribute,being obviously useless and nonsensical on this tag, would be quietly deleted from it.

I have adapted this DTD to a format usable with the W3C validating engine and it canbe invoked by affixing the following declaration to a file:

<!DOCTYPE HTML SYSTEM "https://backiee.wasmer.app/http_www_the-pope_com/html0.h.dtd">
  1. Working KEY Tag Element Example
  2. The Lost Attribute STYLE of DL
  3. The Lost Attribute SOURCE ofBLOCKQUOTE
  4. The Lost AttributesURN and METHODS of A and LINK
  5. The Lost Attribute WIDTH ofPRE

"HTML 1.k" A First Draft Published for HTML 1

There is no official "HTML 1," but if anything could be properlyregarded as at least a semi-official first published draft for an HTML 1, thiswould have to be it. This first published version of HTML was published in themiddle of 1993.For this version of HTML, its reliance upon the previous DTD is obvious, but now somemore useful things have been added. In particular, <BASE> at lastmakes its introduction, an <IMG> tag (imported directly fromMosaic) is also introduced, the TYPE attribute of <A>and <LINK> is at last replaced with the pair of equal butopposite attributes REL and REV, and the STYLEattribute of <DL> finally matures into the COMPACTattribute, but by what appears to be a mistake the COMPACT attributewas omitted from <DL> even as it was added to all of the otherlisting tags, <UL>, <OL>, <DIR>,and <MENU>. One sees this both from the way STYLEwas illustrated in the previous DTD with a value of COMPACT and theway COMPACT is described in this version's DTD, and also from theattached narrative description of this version of HTML. By a similar mistake, thesecond release of HTML 1 nearly a year later will finally affix COMPACTto the <DL> tag, but then deprive the other listing tags ofthis attribute.

In fact, there are a number of differences between the HTML 1 described in thenarrative documentation versus the DTD presented in the latter portion of the samefile. Besides describing COMPACT as an attribute of <DL>it mentions two attributes ALT and ALIGN of<IMG> that the DTD does not support. The narrative would limit<PRE> to taking only certain inline tags where the DTD definesit as accepting all inline tags, including the <KEY> tag whichgoes altogether unmentioned in the narrative. On the other hand, the narrativementions the <HPn> tags in the same manner as it mentions the<XMP>, <LISTING>, and <PLAINTEXT>tags, namely as obsolete, without hinting that only the latter are included in theDTD (though sequestered in a different area with a comment to the effect that theyare obsolete). Finally, the narrative is quite explicit in stating that theN attribute of <NEXTID> cannot take any letters butmust be a number (as indeed was the case in the previous DTD and as observed in allprevious versions), blissfully unaware that the DTD text itself expressly permitsN to be a "NAME" such as Z67.

So which is the "real" first HTML 1 draft? The way I see it, the DTD isthe final arbiter as to what is or is not part of the language. So here is the HTML1 (first draft) set of tags from the DTD, as published in June 1993:

Comments
Document Level Tags
Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
User Input Tags
Listing Tags
Dictionary Tags
Inline Data Type Tags
Inline Formatting Tags
Inserted Object Tags

I have adapted this DTD to a format usable with the W3C validating engine and itcan be invoked by affixing the following declaration to a file:

<!DOCTYPE HTML SYSTEM "https://backiee.wasmer.app/http_www_the-pope_com/html1.k.dtd">
  1. The Lost attribute COMPACT ofUL, OL, DL, DIR, andMENU

"HTML 1.m" A More Advanced HTML 1 Version

By May 18, 1994, a more advanced version of the HTML 1 DTD was posted on theweb, which as seen here,and which also introduces the two level concept seen in HTML 2 where the firstlevel lacks forms and the second has forms. In this form the <P>tag was also transitioning from a separator tag to being a container tag.Indeed, depending upon options selected, it could go either way, with thedefault being that they are container tags. The other interior listing tags(<LI>, <DT>, and <DD>)however all unconditionally become container tags where they had been separatortags from their beginnings up until this point. For backward compatibility, theclosing tags on these listing tags are all optional. The <KEY>tag is going away as something no longer wanted, allowable, but by default notallowable, but at least it was recognized as something more than merely atypographical error for <KBD>. Unlike HTML 2 which onlyhas two selectible options of yes or no each (resulting in four flavors ofHTML 2, this release of HTML 1 has three such yes/no selectible options,resulting in 8 different combinations. To get any combination but the basicdefault however requires additional SGML commands to an HTML<!DOCTYPE> declaration, any of which will result in puttingan unwanted "]]>" showing at the top of the file. Thethree optional flags are:

In addition, the <KEY> and <U> tagscan also be individually enabled, though the default is to exclude them, andthe <NEXTID> can be individually disabled, though thedefault is to include it. The documentation states that the <IMG>element takes "two attributes," lists three attributes (SRC,ALIGN, and ALT), and the DTD lists all three of theseplus a new ISMAP attribute (copied from HTML+) that for the firsttime in HTML enabled image maps. Also introduced are the <BR>and <HR> elements. In addition, a new <STRIKE>element is introduced into the DTD but nothing is said of this in thedocumentation. More can be read about this version of HTML (this one is otherwisepretty well documented) atThis morecomplete HTML Version.

Comments
Document Level Tags
Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
User Input Tags
Listing Tags
Dictionary Tags
Inline Data Type Tags
Inline Formatting Tags
Inserted Object Tags

I have adapted this DTD to a format usable with the W3C validating engine andit can be invoked by affixing the following declaration to a file:

<!DOCTYPE HTML SYSTEM "https://backiee.wasmer.app/http_www_the-pope_com/html1.m.dtd">

It was during this period of HTML that the SGML declaration began to beused as a way to invoke the IETF and Web Consortium HTML validator. Filesfrom this time (and no older) sometimes contain the following:

<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML//EN">

This declaration, once introduced, pointed to the current workingversion of the HTML DTD, whatever it was, and with whatever changes werebeing quietly introduced in the definition. With the introduction of HTML2.0 this began to point to the most default flavor of HTML 2.0, namelythat which is level 2 (permits forms) and not strict (also permits thedepriciated elements). This DTD declaration is still valid to this dayand still points to this particular flavor of HTML 2.0.

HTML+: An Early Attempt to Encompass Everything

Dave Raggett took a different view of writing DTDs than Dan Connolly orTim Berners-Lee, who both seemed to take a rather minimalist approach.Dave's approach was to make the DTD into some sort of "superset" ofall the known tags and attributes currently found out there, whether frompropritary browsers or academia or wherever. So he made an attempt to explorefar ahead of the official draft for HTML 2 already in progress, which hecalled HTML+, as a hint of what was to come later, also published late 1993. In Aprilof 1994, an updated version of HTML+ waspublished. Detail comparisons between the surviving versions of HTML+ andthe Draft of HTML 3.0 (somewhat similar to the comparison listings in thisfile) can be seen here.

Dave Raggett's approach would resurface in the original drafts of HTML3.0 (which were never quite approved), and much of his influence also showsin the final 3.2 version of HTML despite some rather draconion cuts made to it.The problem with it was that the industry was coming up with a whole host ofpresentational features (stylistic commands) where the more academicallydisciplined approach of Connolly and Berners-Lee was that such things shouldbe relegated to some sort of "style sheets," though that concept hadnot been so much as prototyped as yet. The industry, impatient for thedevelopment of style sheets, went ahead and introduced its own tags andattributes at will, and Raggett's approach simply captured that, accidentlyalso serving as a kind of endorsement of them all that Connolly and Berners-Leedid not wish to have given.

HTML 2.0: The Oldest Recognized W3C Version

HTML 2.0 is documented by the Web Consortium in the following document:HypertextMarkup Language - 2.0, dated September 22, 1995. It represented thefirst widely usable form of HTML, and the oldest which the W3C validatingengines can validate without warnings. As it stands, this language comes infour basic flavors, two varieties each of two levels. The first level isthe more stripped-down HTML format, but in fact differing from the secondlevel only in that it tolerates no FORMS tags or attributes. Each of theselevels has a regular and "strict" form, and the strict form prohibitsthe tags that were already depreciated in HTML 1 as "obsolete," namely<NEXTID>, <PLAINTEXT>, <XMP>,and <LISTING>. Apart from those differences, the varietiesof HTML 2.0 scarcely differ.

HTML 2 added relatively little, namely only the <META>tag, and an attribute VERSION to the <HTML>tag, but it also corrected the various problems with the COMPACTattribute, finally assigning it to all the list tags equally, and also fixeda long-running bug with <PLAINTEXT> by making it acontainer tag, with the closing tag optional (i. e. meant to be left off).The <LINK> element at last has the useless NAMEattribute removed from it as one other bit of clean up, but the<DFN> element was eliminated without explanation. The bigdifference that HTML 2 introduced was the use of a strict versioning system(indicated with the new VERSION attribute) that would be used forall versions of HTML from this point onwards. For the first time the variousflavors of the current version of HTML can each be selected without usingclumsy SGML commands. However, the correct place for the version informationhas proved to be in the <!DOCTYPE> declaration, not theVERSION attribute, so the new attribute was depreciated in HTML4.0 and 4.01 and elminated altogether in XHTML 1.0 (but mysteriously resurrectedin XHTML1.1!). This new attribute is not mentioned anywhere in the documentationfor HTML 2.0, but can only be gleaned from inspecting the DTD in detail. Inthe example files called from this file (where acceptible) I have populated thisattribute with its correct value for the file in question.

In HTML 2.0, I identify the following clusters:

Comments
Document Level Tags
Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
User Input Tags
Listing Tags
Dictionary Tags
Inline Data Type Tags
Inline Formatting Tags
Inserted Object Tags

As one can see from the color coding of the above, only a very few of thespecific items codified in HTML 2 were either listed as depreciated or quietlyslated for deletion, such that most of them could not be used in the strictversions of HTML 2, and others would not be seen in any future version ofHTML.

  1. The Introduction of the ALIGN Attributeof IMG and INPUT
  2. The Extraordinary Flexibility of METAand LINK
  3. The HTML VERSIONAttribute (distinct from the <!DOCTYPE> declaration).

HTML 2.0 can be validated with any of the following:

<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0 Level 1//EN"><!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0 Strict Level 1//EN"><!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN"><!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0 Strict//EN">

It should be obvious which of these is for level 1 or level 2 andwhich are for strict or non-strict.

HTML 3.2: A Much Richer Standard, and a Road Not Taken

HTML 3.2 is documented by the Web Consortium in the following document:HTML 3.2 ReferenceSpecification, dated January 14, 1997. This next phase of the developmentof HTML represents an interesting side path taken as a result of industrypressure to which the IETF and W3C barely consented (in fact, IETF bowed outof any further HTML considerations before any version of HTML past 2.0 couldbe published. The industry had found it expedient to hang all manner ofpresentational bells and whistles off the existing HTML tags, and many of themhad already gained considerable industry-wide support, even from competingBrowser and other vendors. In what is almost a capitulation to thefait accompli of all these presentational bells andwhistles, the W3C most grudgingly, and with much delay, finally released their3.2 standard. There had been a 3.0standard beingproposed, but it was never published as anything more than a draft, and itfeatured only all the more such vendor-specific bells and whistles.

In HTML 3.2: I identify the following clusters:

Comments
Document Level Tags
Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
User Input Tags
Listing Tags
Dictionary Tags
Inline Data Type Tags
Inline Formatting Tags
Inserted Object Tags
Table Tags

So one can see the addition of many such universally accepted extensionsto HTML, some of which, such as MAP or TABLE, wouldendure, but many of which would be soon shifted over to style sheets, oncethat technology could be itself settled upon. As it is, the implementationof Style sheets as yet still fails to be anywhere near as consistent amonguser agents as the implementations of many of the above presentationalattributes and tags. In addition, many proprietary extensions weresupported only by one vendor, and competing extensions supported bythe competing vendor, that many commonly used tags at the time were notapproved within any W3C standard, though a few such would finally make itinto HTML 4.

  1. Working APPLET Tag ElementExample
  2. Working TYPE,START, and VALUE Listing AttributeExamples
  3. Working BGCOLORand BACKGROUND Attributes of BODY, TABLE,TR, TH, and TD Tag ElementExamples
  4. Working FONT,BASEFONT, U, and STRIKE Tag ElementExamples
  5. Working ALIGN,NOSHADE, SIZE, and WIDTH Attributesof HR Tag Element Examples
  6. The Addition of the ALIGN Attribute ofmany Tags/Elements, and the CENTER Tag Element Examples
  7. Working CLEAR Attribute ofBR Tag Element Examples
  8. Theuse of STYLE and SCRIPT in HTML 3.2
  9. Working NOWRAP Attribute ofTH and TD Tag Element Examples
  10. Remaining Sizing and Spacing Attributes of HTML 3.2
  11. Netscape Proprietary Extensions to HTML:BLINK
  12. Netscape ProprietaryExtensions to HTML: LAYER
  13. Netscape Proprietary Extensions to HTML: ILAYER
  14. Netscape Proprietary Extensions to HTML:Javascript Stylesheets
  15. MicrosoftProprietary Extensions to HTML: MARQUEE
  16. Microsoft Proprietary Extensions to HTML:COMMENT
  17. Microsoft ProprietaryExtensions to HTML: EMBED
  18. MicrosoftProprietary Extensions to HTML: BGSOUND
  19. Microsoft Proprietary Extensions to HTML: DYNSRC,CONTROLS, START, and LOOP Attributes ofIMG
  20. Microsoft ProprietaryExtensions to HTML: VBScript

HTML 3.2 can be validated with the following:

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2 Final//EN">

HTML 4.01: A Mature and Stable Markup Language

HTML 4.01 is documented by the Web Consortium in the following document:HTML 4.01 Specification, datedDecember 24, 1999. In HTML 4.01, many of the presentational features introducedin HTML 3.2 are depreciated, generally in favor of style sheets, but occasionallyscripting languages as well. Nevertheless, several such proprietary extensionsknown at the time of HTML 3.2 were introduced in HTML 4.01 as depreciatedelements, just so they can be used properly, if used at all. HTML 4.01 Transitionalis the most broad expression of HTML, excluding only a small handful of the veryoldest tags, and at the new end omitting only the XHTML-exclusive Ruby text feature.HTML 4.01 is preceded by an almost identical version (only a few minor typographicalerrors had been fixed) known as HTML4.0. HTML 4.0 (and 4.01) introduced a new <IFRAME> elementwhich is not spoken of as depreciated and yet the Strict forms of this version ofHTML do not admit this element. Perhaps this was accidently done because the newlyintroduced <FRAMESET> element, and other framing elements andattributes, were intentionally depreciated (not included in the Strict versions), and<IFRAME> just got picked up along the way due to the similarityof its name. HTML 4.01 possesses several small groups of attributes which it hangsoff of nearly every different tag. To save display space, I am compacting the eventscategory of attribute under the one heading %EVENTS to replace thelengthy list of onClick, onDblClick, onMouseDown, onMouseUp, onMouseOver, onMouseMove,onMouseOut, onKeyPress, onKeyDown, and onKeyUp. In HTML 4.01, I identify thefollowing clusters:

Comments
Document Level Tags
Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
User Input Tags
Listing Tags
Dictionary Tags
Document Update Tags
Inline Data Type Tags
Inline Formatting Tags
Inserted Object Tags
Table Tags
Frames Tags
  1. Attribute PROFILE of HEAD
  2. Handling Different Languages in HTML
  3. Working DATASRC, DATAFLD,DATAFORMATAS, and DATAPAGESIZE Attributesof TABLE and other Tags Elements Examples
  4. ISMAP Attribute of Forms INPUT Tag Element
  5. Remaining Sizing, Spacing, and Coloring Attributes ofHTML 4.0 and 4.01

HTML 4.0 and 4.01 can be validated with the following:

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"><!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Frameset//EN"><!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0//EN"><!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"><!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Frameset//EN"><!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN">

XHTML 1.0: Changing the Basis from SGML to XML

XHTML 1.0 is documented by the Web Consortium in the following document:XHTML™ 1.0 The Extensible HyperTextMarkup Language, dated August 01, 2002. XHTML 1.0 possesses nearly thesame exact list of tags and attributes as HTML 4.01, since it is little morethan a conversion of HTML 4.01 into an XML-based language. Nevertheless, itdoes introduce some changes from HTML, most notably the introduction of thexml:lang, xmlns, and xml:space attributes,and the complete elimination of the version attribute. It also madethe use of lower case tags and attributes mandatory, which I show here as listingall tags and attributes (even those expired) as lower case. Before they werecase-insensitive, but upper case was most typical so as to make the tags standout from the text as much as possible. As I did above with HTML 4.01, I amcompacting the events category of attribute under the one heading%events to replace the lengthy list of onclick, ondblclick,onmousedown, onmouseup, onmouseover, onmousemove, onmouseout, onkeypress,onkeydown, and onkeyup. In XHTML 1.0, I identify the following clusters:

Comments
Document Level Tags
Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
User Input Tags
Listing Tags
Dictionary Tags
Document Update Tags
Inline Data Type Tags
Inline Formatting Tags
Inserted Object Tags
Table Tags
Frames Tags

XHTML 1.0 can be validated with the following:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd"><!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">

XHTML 1.1: Leaving the Past Behind

XHTML 1.1 is documented by the Web Consortium in the following document:XHTML™ 1.1 - Module-basedXHTML, a working draft dated February 16, 2007. XHTML 1.1 possesses onlythe same list as XHTML 1.0 Strict, plus it depreciates yet some more items andadds a few items. By far, the most notable addition is the new Ruby textfeature, which allows a small-type text to be affixed to regular-sized text.As I did above with HTML 4.01, I am compacting the events category of attributeunder the one heading %events to replace the lengthy list of onclick,ondblclick, onmousedown, onmouseup, onmouseover, onmousemove, onmouseout,onkeypress, onkeydown, and onkeyup. In XHTML 1.1, I identify the followingclusters (to be consistent with my previous listings; these only loosely coincidewith the various XML modules selected for "normal" XHTML 1.1, but thereare some minor differences):

NOTE: The following listing has not yet been validated!

Comments
Document Level Tags
Document Head Tags
Block Level Body Tags
Raw Text Tags
Hyper Link Tags
User Input Tags
Listing Tags
Dictionary Tags
Document Update Tags
Inline Data Type Tags
Inline Formatting Tags
Inserted Object Tags
Table Tags
Frames Tags
  1. Working rubytxt Tag ElementExamples

XHTML 1.1 can be validated with the following:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">

HTML 5.0 and XHTML 2.0: What's to Come

XHTML 2.0 is documented by the Web Consortium in the following document:XHTML™ 2.0, a working draftdated July 26, 2007. However, protest from the Joe Sixpack programmers (andothers) is pressuring the Web Consortium to return to advancing regular SGML-basedHTML, at least as a parallel effort to their development of XHTML. This is notonly driven by the relative ease by which HTML can be programmed as opposed tothe severe strictness of XHTML, but also by the "spagetti-dtd" nature ofXHTML 1.1's document definition. Suddenly there is no longer simply one big filethat does it all, but instead quite a swelter of little files that are next toimpossible to get through so as to ascertain the actual contents of the language.So, in response to the call to resume development of HTML, a sort of draft orforum for HTML 5.0 is made available hereAs one can see from the working drafts, there is much about HTML 5.0 and XHTML 2.0which are still in a state of flux, and much that remains to be written asyet. Furthermore, it is also impossible to validate a document purporting tobe an XHTML 2.0 document at this time, and only some of the more trivial HTML 5.0documents can be correctly validated, since the language itself is still underdevelopment and furthermore basing itself on neither XML nor SGML, so only a kindof state-machine validator is available for beta testing.

This file, "lostHTML.htm," is XHTML 1.1compliant.
The XHTML 1.0 Strict counterpart to this file, lostHTML_html is XHTML 1.0Strict compliant.
The Source and Contents of the oldest surviving HTML Filesfile "htmlsource0.a_html" is HTML 4.01 Strictcompliant.
The Source and Contents of a sampling of the earliest documented HTMLfile "htmlsource0.c_html" is HTML 4.01 Strictcompliant.
The Discussions for Future Directions for HTMLfile "htmldirections_html" is HTML 4.01 Strictcompliant.
The Source and Contents of a sampling of the earliest SGML-compliant HTMLfile "htmlsource0.d_html" is HTML 4.01 Strictcompliant.
The Crazy SGML Stuff demonstration filefile "sgmlcraz_html" is HTML 4.01 Strictcompliant,but with 5 warnings.

Next Level Up