>
> > I think it's always said that, hasn't it? In any case, thanks for
> > finding it: if we are describing practice, then it ought to say
> >
> > ...which specifies if the graphic is to be aligned with the TOP,
> > MIDDLE, or BOTTOM of the adjacent text.
> >
> > because it's usually the image which moves, not the text, isn't it?
>
> Yes, usually the image is moved to align with the top, bottom or middle
> of the text (or image) that is just before it in the line.
Excuse me, but I have yet to see a browser which moves the image up and
overlays text on the previous line which is what this description
implies. Clearly the vertical space required for all elements placed
on a line is figured out first. Then the relative positioning is
adjusted based on defaults, align= etc. What is attempting to be
described is the meaning of the align attribute. Therefore I would
propose:
...which requests apparent alignment of the adjacent text
and the graphic as middle to middle, top to top or bottom
to bottom.
Anything more than that should be left to the UA implementors. I see
no reason why an implementation should be discouraged from implementing
middle to mean that element text prior the <img> shouldn't flow to
multiple lines with typographically weighted centering and a similar
approach might end up with a different number of lines following
the <img>, etc. Until we have the HTML3 features for specifing how
text flows about figures, tables, etc., current practice on the
part of information publishers is going to be to tread very
carefully. Current practice of browsers breaks today when the text
doesn't fit on the line with the image. So publishers knowing that
will tend to keep the text short. Hence, there should be no breakage
with a purposefully general statement such as I have attempted to
make.
Dave Morris