HyperText Mark-up Language Quick Reference, December 1994

This is yet another HTML quick reference, containing advice on how to write correct HTML, as well as practical tips. It is mainly based on version 1.22 of the HTML 2.0 standard (including some HTML3 extensions which are already implemented by a number of browsers; the extensions are all clearly labeled as such). Some material is from another quick reference by Tom Fine. This is not necessarily a good guide for absolute beginners.


The HTML language represents hypertext data, for use as part of the World-wide Web. HTML is one specific language defined using the general SGML meta-language. HTTP is a transport protocol, used to deliver HTML documents (as well as other types of files) over networks.


<tagname attribute=value attribute="Value"> contained stuff </closingtag>

A "tag" is everything between the `<' and `>' characters. The tag name should come directly after the `<' character, with no intervening whitespace. Tag names and tag attributes are case-insensitive, as are the values of certain tag attributes as well. If an attribute value contains whitespace, or any characters other than a-z, A-Z, `.' or `-' it should be quoted. For this reason, most URL's should be quoted (the fact that some implementations may tamper with the alphabetic case of unquoted attribute values means that it is good style to quote all URL's). Some attributes (such as COMPACT) do not need a value.

An "element" is made up of the opening tag, its matching closing tag, and everything that contained between the two (which can include other tags, and also text which is not part of any tag): <X>Stuff In Element</X>. The closing tags for some elements are optional (as noted below), and some tags can not have a corresponding closing tag (namely, <BR>, <HR>, <IMG>, <INPUT>, the non-<TITLE> tags in <HEAD>...</HEAD>, and the SGML pseudo-tags <!DOCTYPE> and <!-- -->).

Details of text formatting in the HTML source (such as the position of linebreaks) are not preserved when the document is displayed, and extra whitespace is ignored.

Overall Document Tags:

<!DOCTYPE ...>
SGML declaration; if it is used, it comes first in the file. (If you don't know what this is, don't worry about it.)
<HTML> ... </HTML>
Encloses the entire document (except <!DOCTYPE>) and identifies it as HTML. The optional VERSION attribute specifies the HTML version used.
<HEAD> ... </HEAD>
Encloses the header (<TITLE>, <LINK>, etc.).
<BODY> ... </BODY>
Encloses the body of the document.

All other tags besides these, and all text which is not part of a tag, should be contained within a <HEAD>...</HEAD> or <BODY>...</BODY> element, which should be in turn contained within the overall <HTML>...</HTML>.

Head Tags:

<TITLE> ... </TITLE>
The title of the document; should not contain any other tags. A title is OBLIGATORY in <HEAD>...</HEAD>! There should be no non-tag text in <HEAD>...</HEAD> except that which is contained in <TITLE>...</TITLE>. The title is used to index the document by Web searchers, so that it should not be something which is cryptic when out of context, like "Intro". (See also <H1>...</H1> below.)
Specifies general relationships of this document to other resources. The type of relationship is described by a REL= or REV= attribute (other attributes are URN=, TITLE=, and METHODS=). This is not generally implemented yet, except for <LINK REV="made" HREF="mailto:...">, used to specify the e-mail address of the author.
Specifies context-independent URL of current file.
<NEXTID N="...">
Next anchor name to use (for HTML editors).
Provides HTTP header info (other attributes are NAME= and HTTP-EQUIV=).
Document is searchable index. (This tag can also appear in <BODY>...</BODY> element.)

Body tags:

High-level tags:

<Hn> ... </Hn>
Section headings; "n" ranges from 1 (highest-level) to 6 (least important); <H4>-<H6> are too small to be usable in the default configuration of some versions of NCSA Mosaic. Since the TITLE in the HEAD element is displayed on the window bar (and should be context-independent), the <H1>...</H1> element is generally used for the actual within-document title.
Encloses a block of text that is a quote.
Information about the author and the document itself (such as copyright, sources, last update, acknowledgements, etc.). Shouldn't include lists or high-level tags (except <P>). Often displayed as italic.
Horizontal line (pseudo page-break).
<PRE> ... </PRE>
Encloses block of text to be shown verbatim in a fixed-width font (whitespace is significant). This is the only way to do columns or aligned tables in HTML 2.0. The WIDTH= attribute gives a display hint to browsers (the default is WIDTH=80). A <PRE>...</PRE> element cannot contain any list or high-level tags except <HR>.
Begin a paragraph (the closing </P> tag is optional). Cannot contain lists, or any of the above tags. Use attributes ALIGN=CENTER or ALIGN=RIGHT to control text position (this is not yet a part of standard HTML 2.0).

These high-level elements all imply both a preceding and a following paragraph break (except after the optional </P> tag).


<UL><LI>... </UL>
Unordered list.
<OL><LI>... </OL>
Ordered list.
<MENU><LI>... </MENU>
Menu list (for brief items; not much used -- you can also try <UL COMPACT> or <OL COMPACT>).
<DIR><LI>... </DIR>
Directory list (should be multi-column, but isn't in most implementations).
Begins each item in the above lists. An inline image can be used as a custom bullet (preceding the list item) using the attribute BULLET="URL". (This is not yet a part of standard HTML 2.0.)
<DL><DT>... <DD>... </DL>
Definition list. Can also be useful for writing dialogue (as in a play). Use <DL COMPACT> for tighter rendering. There does not actually have to be a <DD> for each <DT> or vice versa.
Begins each item title in DL.
Begins each item definition in DL.

The list item closing tags </LI>, </DT>, and </DD> are optional.

Lists can be nested (i.e. included in an <LI>...</LI> item in a <UL>...</UL> or a <OL>...</OL> list, or inside a <DD>...</DD> item in a <DL>...<DL> list). List items are not supposed to directly contain <H1>-<H6> headings, <HR>, or <ADDRESS> (though <LI> and <DD> elements can contain a <BLOCKQUOTE> or <FORM> which itself includes them).

Hypertext Anchors/Links:

<A NAME="..." HREF="URL"> ... </A>
Creates a link (HREF=) or anchor (NAME=) or both. (Less commonly used attributes are URN=, REL=, REV=, TITLE=, and METHODS=.) Non-HTML resource types referenced in <A HREF="..."> links can be displayed by external viewers. An anchor element MUST contain something other than whitespace, or it won't work on many browsers.

It is better if text contained within a link element is not something relatively meaningless like <A...>Click Here</A>, but rather something which describes what the link is pointing to: <A...>Chelsea's cat Socks</A>. (Remember that not everybody is using a mouse anyway, so the word "Select" is preferable to "Click".)

Anchors/links CANNOT BE NESTED, directly or indirectly, so that even code like <A...>...<X>...<A...>... </A>...</X>...</A> is forbidden. (In the upcoming HTML3 language, the attribute ID="...", which will be able to be used with most tags, will replace <A NAME="...">, so that almost any element will be able to be the target of a link.)

Inline images:

<IMG SRC="URL" ALT="...">
Inserts an image from the URL as part of the surrounding text flow (if any); GIF 87a (.gif) and X Bitmap (.xbm) formats are supported. (JPEG's are supported by Netscape.)
These attributes control the placement of short captions alongside an image (but will probably not do what you expect in the middle of text).
Specify the width and height of the image in pixels. Greatly speed display of document in Netscape (not yet part of standard HTML 2.0).
The image is a clickable imagemap.

Be sure to specify meaningful text in the ALT attribute value (for use in non-graphic environments), especially if the image is in a link. If the image is purely decorative, use ALT="" to avoid annoying "[IMAGE]" clutter in Lynx.

Using too many and too large inline bitmaps can be very inconsiderate, especially on your home page and other pages that are linked to from outside, (unless they are publicized as picture galleries). Many people are using 14.4k modems, and it is particularly frustrating when with no advance warning you have to wait for a lot of big .GIF's to load -- before you're even able to decide whether or not there is actually anything of interest on the page. In any case, inline images will often be shown with few colors (only 50 in some versions of Mosaic), whereas external images will be shown with the maximum available number of colors -- so it is best to use a small sample (thumbnail) as a link to the full size image.

Text flow:

Forces a line break

Logical styles:

<EM> ... </EM>
Emphasized (often rendered as italic).
Strong emphasis (often rendered bold).
<CITE> ... </CITE>
Citation of book, article, movie, etc. (often rendered italic).
<CODE> ... </CODE>
Piece of computer source code (often rendered in fixed-width font).
<KBD> ... </KBD>
Example of keyboard entry (user input).
<SAMP> ... </SAMP>
Literal characters (e.g. computer output).
<VAR> ... </VAR>
Name of variable (often rendered as italic).
<DFN> ... </DFN>
Word to be introduced/defined (not yet part of standard HTML 2.0).


<B> ... </B>
Bold font.
<I> ... </I>
Italic font.
<TT> ... </TT>
Typewriter (fixed-width) font.
<U> ... </U>
Underlined (not yet part of standard HTML 2.0; can also create confusion with links, which are rendered as underlined on many browsers).

It is preferable to use logical styles rather than hard-wired fonts (bold, italic, etc. may not be available in non-graphical environments, anyway). Styles and fonts are NOT guaranteed to be rendered cumulatively (i.e. <B><I>Text</I></B> may look the same as plain <I>Text</I>, and the italic text in <H1>RomanText <I>ItalicText</I></H1> may not be the appropriate size for a H1 heading).

The logical style, font, and link/anchor elements generally can contain only each other (and <IMG> and <BR>), and not lists and high-level tags. The headings <H1>-<H6>, <DT>...</DT> in a <DL>...</DL> list, and <LI>...</LI> in MENU or DIR can also contain only these tags. It is best not to have whitespace after an opening tag of a style, font, or anchor element, or before a closing tag (i.e. <B>Text</B> is preferable to <B> Text </B>); such whitespace produces displeasing visual results on some browsers.

Special Characters (Entities):

&amp; or &#38;
&lt; or &#60;
&gt; or &#62;

These three characters should be escaped with the above ampersand entities everywhere in a document where they are not intended to be used with their HTML meanings. Other entities (such as "&eacute;" etc.) are available to encode the alphabetic characters in positions 192-255 of the ISO 8859-1 Latin 1 character set for European languages. Numeric entities can be used for characters in the range 160-191 with some hope of success (such as &#169; for the copyright symbol, since not all browsers understand &copy;). Not all browsers understand &nbsp; or even treat &#160; as a space -- a safe alternative is &#32; (but this will not act as non-breaking on most browsers). The range 127-159 is undefined in ISO 8859-1, and should not be used. A double-quote character must be escaped as &quot; or &#34; inside an attribute value. Characters in URL's are best escaped with %-hex-digits (e.g. %2B for "&").


<!-- comments go here -- >
The stuff in such a tag is ignored. The final "--" marks the end of the comment. Theoretically, a comment can include other HTML tags, but you're much wiser NOT doing this, since many implementations don't support it. Some implementations restrict comments to a single line.


General form:

Where protocol is one of http, gopher, ftp, file, telnet, wais, news, mailto, etc. The "#anchor" is optional, and ":port" defaults to 80 if left out.

A fully absolute URL contains a protocol prefix, and a full hostname for external DNS resolution.

Absolute URL:

A URL can be relative in several ways:

Server-relative URL:
Document-relative URL:

(Uses the protocol, host, and port of the current document.)

Document-internal URL:

URL's which are document-relative, but specify something outside the current directory (i.e. http URL's which contain a `/' character, but do not start with a `/' character, after the optional protocol prefix) can sometimes confuse browsers (especially relative URL's that start with "../" -- in general, ".." will be interpreted in terms of the logical Web file system, rather than the physical file system).


To implement forms (or <ISINDEX> or <IMG ISMAP>) you need special HTTP-server stuff outside your HTML file.

<FORM ...> ... </FORM>
Encloses the entire form.
The URL to use when the form is complete
Mime type of representation of form data
<INPUT ...>
Some type of input field.
Types are text, password, checkbox, radio, submit, reset, image, hidden
Name of the field
Value of button (label for submit and reset)
URL of inline image (image)
This item selected by default (checkbox/radio)
Displayed field width, in characters
Maximum field width, in characters
Image alignment (image)
<SELECT ...><OPTION ...> ... </SELECT>
A list of items to select.
Name of the field
Use scrollable list with SIZE # options shown
Multiple selections allowed
<OPTION ...>
Precedes each item in the option list. The closing </OPTION> tag is optional.
This option is selected by default
A multiline text field. The enclosed text is the default value displayed in the field.
Name of the field
Number of rows in the field
Number of columns in the field

<FORM>...</FORM> is a high-level element, and so should not be contained inside a heading, <ADDRESS>...</ADDRESS>, <PRE>...</PRE>, <P>, <DT>...</DT>, style, font, or anchor element (or <LI>...</LI> in MENU or DIR). <INPUT>, <SELECT>, and <TEXTAREA> should only be contained within a form. Forms cannot be nested.