Comments on: Gradients, Fills, and Shadows, Oh My

By: Jon Peltier

Jon Peltier — Sat, 16 Aug 2008 22:49:38 +0000

Alexa - Microsoft removed that capability from the user interface (the ribbon and all that), but it's still accessible to VBA. Andy Pope has written a free add-in for Excel 2007 that provides a dialog to apply these fill patterns to Excel 2007. Read about it in Chart Pattern Fills Add-In.

By: Alexa Class

Alexa Class — Sat, 16 Aug 2008 20:46:07 +0000

Is there a way to add diagonal lines to a section of bars or a bar through Excel 2007. I only find gradients/ color options and am trying to publish scientific info where they require diagonal black and white lines instead of gradients. Help very appreciated

By: Jon Peltier

Jon Peltier — Mon, 11 Aug 2008 01:52:58 +0000

John - I've described the rationale behind my decision to sort the date axis categories by date in Choice of Category Axis Order.

By: Jon Peltier

Jon Peltier — Sun, 10 Aug 2008 23:55:14 +0000

John -

I’ve followed up a bit more, and an upcoming post will have a bit more detail. It would seem that Chip’s source for the calculation is not correct.

The order of categories makes sense either way, depending on what you’re interested in. I am interested in the distribution across the range of dates, and I describe it further in the upcoming post. A pareto like yours will give different results, at least a different order of categories, depending on whether you sample 200, or 300, or 500 years. The distribution in date order will look much the same independent of sample size.

By: John Walkenbach

John Walkenbach — Sun, 10 Aug 2008 21:47:49 +0000

OK, I concede the first point. It was a moment of insanity. Excel 2007 just made it too easy to spoil a good chart. I throw myself on the mercy of the court.

I see your point regarding data order vs. frequency order, but I prefer my method.

According to several web sites I visited, Easter 2079 is on April 23. Chip’s VBA function returns that date, but his formulas return April 16.

By: Jon Peltier

Jon Peltier — Sun, 10 Aug 2008 20:15:57 +0000

Jon – Despite the color scheme, the clustered bar chart is not very useful for showing the differences, especially if there are relatively feew differences. I wouldn’t use error bars, because you’d have to pick one series to apply them to, which would discount the importance of the other. I thought the line chart showed the differences best.

By: Jon Spring

Jon Spring — Sun, 10 Aug 2008 20:04:01 +0000

Another easy way to show the discrepancy would be to add error bars to your basic date-order chart above. The important story is that only those two dates show any difference. The problem with the paired-bar chart is that the color clash makes you look through every date to find the story, not just the two meaningful dates.