Steve

> > Your solution (whilst very welcome) doesn't seen semantic, as the 
> > container DIVs in the nest have no semantic meaning.
> 
> Obviously you know more about it than I do, 

Your compliment reflects better on you than me ;)

> but there is no way for me 
> to know that without knowing the content.  :)  :)  There certainly 
> _could_ be a page wherein that structure would be perfectly logical 
> semantically.  That was why I was asking to see the content, as 
> semantics are by definition determined by the information and language 
> being used.

I see your point, however... The reason why I stated that the container DIVs have no semantic meaning is that each of the boxes is a logical division of the page. To associate each row of boxes together to form another logical division of the page makes an assumption that is invalid, given the fact that the developer (me) has no control over the content of each of these boxes.

> > In addition, the grid structure will not be maintained if any content 
> > stretches a single cell horizontally.
> 
> Again, not sure what you're doing, but it sounds almost like what 
> you're doing truly is tabular data, in which case a table _should_ be 

I wouldn't think of it as tabular data; mainly because there is no required link or shared definition between any of the data contained within the boxes.

> used.  Sometimes seeing it visually can suggest non-obvious ways of 
> structuring things. For example, is there actual text in all of these 

Whatever the marketroids (who apparently "own" the websites) care to place in them.

> cells or are some of them decorative elements?  Is something 
> essentially a sidebar?  And so forth and so on.

"Possibly" and "no" respectively.

> Another possibility would be to use divs with wrap set in certain ways 
> that would essentially emulate frames?  (I forget the CSS off the top 
> of me head)

A combination of this and your previous suggestion would *almost* cut it to my satisfaction. The only two problems are the previously discussed issue of logical page divisions, and the fact that the design needs to adjust itself automatically if one of the cell widths is forced out. The only time that this will be a issue is when a link like "Next" translates in Gertman to "Sie Drickerzeinvarswolder Giesflekken" and must not wrap. In this case, the page should stretch but retain the integrity of the grid.

> I guess I'm having a hard time imagining non-tabular data that would 
> absolutely require  such a layout.  I'm very curious all of a sudden.  
> :)

I'm afraid it's really not as exciting as it may seem :) The only reason that I'm not more forthcoming about content is that there won't be any until I've completed the template. The design must be content non-specific, and must consisit of the grid I posted earlier.

> I'm not so opposed to using a table for basic layout if it can't be 
> avoided -- the problems generally come into play more when a design 
> involves table nested five deep!  Using one table for the rough layout 
> is probably a misdemeanor, not a felony!

Well, I've used a single table for the basic layout, and other than that I'm pretty pleased with my markup. I'm abandoning the idea of a wholly semantic site for this project, but will continue addressing the problem in my spare time.

Again, thanks for your input.

Cheers!

Regards

Chris Marsh