Some of the more significant changes are listed below:

The mechanism by which emphasised text inherits from the 
enclosing paragraph had a bug. This meant that paragraphs that 
had emphasis from their first character could either lose their 
emphasis, or have the emphasis applied to the whole paragraph, 
rather than to the appropriate region of text. The scheme has 
been simplified and corrected.

New destinations for text in W4W v6 showed up a bug in the 
handling of optionally ignorable destinations. This results in 
spurious text appearing in documents imported from W4W to the S3. 
Such destinations, and all nested destinations within them, are 
now correctly ignored if the optional destination is not 
recognised by the reader code.

Negative line spacings (meaning a fixed line spacing to Microsoft 
Word) are now converted to positive values by the reader. 
Previously they were interpreted as very large line spacings, 
resulting in S3 documents with one line per page.

Incoming negative left margin indents are now truncated to zero 
(they were previously interpreted as large positive indents, 
resulting in text being printed with one character per line).

Added simple interpretation of tables in the reader. Each 
incoming table row is converted to a separate paragraph, with the 
cells separated by Tab characters. Note that this scheme does not 
cope with cells containing multiple paragraphs.

The reader now takes note of the \deff<n> keyword and sets an 
appropriate default font.

The reader now handles the \gutter keyword (simply by increasing 
the left page margin).

The reader now responds to the \sect end of section keyword. The 
response is the same as for \par (end of paragraph).

The additional Macintosh Word keywords \bullet, \endash, \emdash, 
\ldblquote, \rdblquote, \lquote and \rquote are handled on input. 
They are not output since they are not understood by DOS Word; 
this scheme maximises compatibility with both flavours of 
Microsoft Word.