Take care not to confuse the "Added Latin 1" entity set (from an
appendix to the SGML spec, ISO8879) with the Latin 1 character set
(defined by ISO-8859-1).
& and " are not in the "Added Latin 1" entity set -- they're
in the iso-num set ("ISO 8879-1986//ENTITIES Numeric and Special
Graphic//EN"). But the rest of iso-num isn't used in HTML, so the few
definitions for amp, quot, lt, etc. are inlined in html.dtd.
The Added Latin 1 entity set defines a bunch of names for Latin 1
characters. The SGML spec appendix that defines it makes no reference
to the Latin 1 character set (ISO-8859-1). It maps those names to
these thingies called SDATA entities -- system dependent data
entities. I believe the intention is that the SDATA entities are
supposed to be replaced on a per-SGML-system basis. So you might
see TeX version of "ISO 8879-1986//ENTITIES Added Latin 1//EN", with:
<!ENTITY eacute SDATA "\eacute" -- for TeX -->
Since the document character set for HTML includes all the characters
referred to by those names, there's no need to use system-specific
mappings. The entities can be mapped to characters within the document
In response to the same feedback you saw, this set of definitions is
"ISO 8879-1986//ENTITIES Added Latin 1//EN//HTML"