"John Mansfield" wrote in
message ...
Pablo,

I can understand why you might have a component of the variance go
negative but in theory the start and end points should always be
positive. As a result, couldn't you just set your data up similar
to
this and then just create a simple clustered column chart?


Hi John,

Just as a note, this is correct if you are looking at, say, revenue.

However, what about the case where the chart is reconciling, say,
EBIT. The start point might be $1.2m profit, and the end point
perhaps $0.7m loss.

In that situation, we have to be able to cope with a variance bar
crossing the axis I think?

Alan.