In the previous post, the data was widely spaced, so it filled a base 10 log scale with two decades with only a reasonable amount of space above and below the data.

If the data isn’t spread out so nicely, you lose the opportunity to use a regular logarithmic axis. We’ll look at this for Excel 2003 and for Excel 2007, and then we’ll look at native logarithmic axis labeling.

Excel 2003

The data in the table below has a narrow range, from 8 to 12, and the range spans a power of ten.

Here is the data charted using a linear axis.

When we apply a logarithmic scale axis, the data spans across 10, so by default the axis ranges from 1 to 100. The data is squeezed into the middle of the chart.

Since Excel 2003 only permits the axis to begin and end at powers of ten, we’re stuck with this, and the fanciest labeling doesn’t make the data easier to read.

Following the steps in Custom Axis, Y = 1, 2, 4, 8, 16 we can plot the logs of the data on a linear scale, from log(8) = 0.903 to log(12) = 1.079.

We can hide the default labels, add a series with points where we want our custom labels using log(Y) data, and use the Y values as data labels.

Here’s the chart. With such a narrow spread in the data, it’s not immediately apparent that the Y scale isn’t linear, but if you took out your ruler (you still have one of those, right? you’re not completely digital?) you can tell that the span between 8 and 9 is larger than that between 11 and 12.

Excel 2007

As described in Custom Axis, Y = 1, 2, 4, 8, 16, Excel 2007 makes working with log scales a bit easier.

Here is the data plotted on a linear scale.

Here is the same chart, with the scale transformed to a logarithmic scale, using the default base 10. Same as Excel 2003.

Here the log scale has been changed to base 2.

In 2007 we can change the min and max of the log axis to values other than 10 (i.e., other than the base). Here the axis ranges from 8 to 80, still a decade on the base 10 log scale. Instead of the minor ticks being located at multiples of ten (20, 30, 40, 50,…), they are located at multiples of 8 (16, 24, 32, 40,…)

We can also change the maximum so that the axis spans a non-integral number of cycles. Here the maximum is 12 to fit the data. This looks the same using base 10 or base 2.

We don’t get labels other than at the minimum because the axis spans less than a power of base 10 or of base 2, but we can use the same protocol as above to add points with data labels. The advantage Excel 2007 has over 2003 is that we can use the actual values to locate the points, and we can simply use the Y value data label option. In Excel 2003 we had to use the log of the Y positions, and either manually edit the labels, or use a third-party add-in. (Even though the add-in is great, it still is something extra required.)

Finished. Again, just like Excel 2003, only easier.

Log Axis Labeling

Below are a few variations on a log axis that spans six cycles. These were produced in Excel 2003, but they would be the same in 2007. The scales show both major and minor tick marks, which are available for any base except for 2.

The first scale runs from 1 to 1,000,000, and the second from 1/1,000,000 to 1. The third and fourth run from 1 to 1,000,000 like the first, but instead of the default major unit (major tick spacing) of 10, the third has a major unit of 100 and the fourth of 1000. The minor tick spacing is ignored.

If you use the default major unit, minor ticks are placed at multiples of the number at the start of the cycle. In the Excel 2007 chart above that used a base 10 log scale and ranged from 8 to 80, the minor ticks were at multiples of 8 (16, 24, 32, 40,…). In our first axis below, the bottom of the decade is at 1, so minor ticks are at multiples of 1 (2, 3, 4, 5,…). The the second axis, the bottom of the decade is at 0.1, so minor ticks are at 0.2, 0.3, 0.4, and so forth.

In 2003, when the non-default major unit is used, the space between major tick marks is divided up as if it were a single cycle (not two or three. Instead of the ticks being located at 2, 3, 4, etc, for a major unit of two cycles, the minor ticks are located at the square of the one-cycle values, that is, at 4, 9, 16, etc. For a major unit of three cycles, the minor ticks are located at the cube of the one-cycle values, that is, at 8, 27, 64, etc. Interesting.

Excel 2007 does this a bit differently, and for a two-cycle major unit, it makes sense. The first pair of axes below show a scale of 1 to 100, first with the major unit defined by the default one cycle, then by two cycles. I’ve added labels to many of the minor ticks to help illustrate this behavior. The two cycle axis has the first minor tick where the first cycle ends and the next cycle starts, then the rest of the ticks are where they were defined for the second cycle. This puts the ticks at the very logical values of 10, 20, 30, etc.

With larger spans between major ticks, Excel 2007 seems to get confused. The second pair of axes shows three cycles with major units of one and three cycles. The three-cycle axis places minor ticks where the two-axis tick did in the first instance, at multiples of 10 between 10 and 90. There is no tick at 100, nor any between 100 and 1000. Not shown is a four-cycle axis. This puts minor ticks at multiples of 10 between 10 and 90, and at multiples of 1000 between 1000 and 9000, but skips the decade between 100 and 1000.

– - –

This is pretty obscure. It doesn’t make sense to use multiple cycles as the major unit. In a two-cycle major unit, Excel 2007 uses more logical minor tick spacing, while the minor tick spacing is broken for larger major units.  Excel 2003 uses a consistent minor tick spacing which is logical in its own way, but not readily understandable.

I first wrote up that tutorial back when I was using Excel 97, and the protocol remained the same for Excel versions through Excel 2003. The protocol is pretty much the same in Excel 2007, too, but the rearrangement of the interface into the Office 2007 ribbon has made the old instructions difficult to follow.

Let’s use the same data as in the previous example (with recent dates). This is the data for the main series of values in the chart. The average and standard deviation for the values in B5:B21 are displayed in B1 and B2.

The data for the bands is given below. We’ll use stacked area type series. The value in E2 is three sigma below the mean, so the formula in E2 is =$B$1-3*$B$2. The thickness of each remaining band is one standard deviation, so the formula in F2:K2 is =$B$2. The formula in E3 is =E2, and this is copied and pasted into F2:K2.

The data in A4:B21 is used to make a simple line chart. The Y axis min and max limits have been set to fixed  values of 0 and 22, so they don’t change as we’re adding and formatting the rest of the series.

Copy D1:K3, select the chart, and use Paste Special to add the data as new series, with data in columns, categories in the first column (but don’t replace existing categories), and series names in the first row.

These series are added as line chart type series, but we can’t see them (In 2003 we could, so the next step was easier). The values 0 and 1 on the time scale axis are plotted off scale, so we will move the series to the secondary axis. We can’t click on the series to format them, so we have to use the Chart Element control to select them. This control is present at the left edge of the Layout and Format tabs, the last two Chart Tools tabs.

All of the chart elements are listed in this dropdown control. You can select the hidden series using this control.

One by one, select one of these new series, press Ctrl+1 to open the Format Series dialog, and on the first tab you see, select the Secondary Axis option. The line chart series are now visible. The lower series shows the -3 SD data The upper series is really all of the other added series, which coincide because the Y values are all the same.

When Excel assigns a series to the secondary axis, often it only adds a secondary value (Y) axis, which we see at the right edge of the chart. We need the secondary category (X) axis to make the background work. On the Chart Tools – Layout tab, click the Axes dropdown (near the middle of the tab), select Secondary Horizontal Axis, and choose the Show Left to Right Axis option. (Why does it feel like they’ve dumbed it down a little too far?)

One of the changes in Excel 2007 that decimates productivity and efficiency is that you cannot use the F4 key (shortcut for Repeat Last Action) here. In 2003 you could assign one series to the secondary axis, then select another series using the arrow keys and press F4 to quickly assign is to the secondary axis.

We want the 0 and 1 of this secondary axis to extend to the edges of the chart, but we have to wait until after the next step.

We also want to convert the line chart series into stacked area chart series. One by one, select each added series by right clicking, and choose Chart Type from the pop up menu. Select the Stacked Area subtype of the Area Chart type. Do this seven times to get the stacked colored bands.

Here is another situation in which you cannot use the F4 Repeat Last Action shortcut. In 2003 you could change the chart type of one series, then select another series using the arrow keys and press F4 to quickly change it to the same chart type.

Now we can fix the secondary category (X) axis. Select the axis, press Ctrl+1 to open the Format Axis dialog, and at the bottom of the first tab you see, select the On Tick Marks option for Position Axis.

Now let’s clean up the axes. Format the secondary category (X) axis so that no parts of it are visible: choose None for major and minor tick mark type and for axis labels, all on the first tab, Axis Options. On the Line Color tab, choose No Line. Select the secondary value (Y) axis and press Delete. This forces the stacked area series to use the primary value (Y) axis. Format the dates in the primary category (X) axis so they are easier to read.

Those colors have to go. Here is the chart with stacked area series fills formatted with the Standard Colors in the Chart Tools – Format tab’s Shape Fill dropdown. The lowest area series is formatted with no fill, and the plot area is filled with red.

Those are a little too bold, so use the More Fill Colors option to tone down the fills. I used the Standard Colors as a basis, then on the Custom tab of the Colors dialog, switched to the HSL color model, and changed the Luminosity setting to a higher number in the range of 180 to 200.

After my earlier article about chart colors, I used a ColorBrewer-derived palette in Excel 2003. At this point, though, I’ve decided I’m not terribly fond of this palette: the colors are fine together, but they are not the “pure” colors I’d like. The red has an orange tinge, the green is shaded somewhat towards blue, the blue is a bit purplish, etc.

So I’ve changed palettes to be more in line with the colors that one would expect to find, as set forth by Stephen Few.

Here are the default Excel palette, my previous palette, and my current palette. Notable changes are the evolution from Excel’s unattractive default charting colors in the bottom two rows of the palette, to the ColorBrewer derived colors, to the Few inspired colors. In the earlier custom palette (center) I also lightened up some grays in the right column of the palette, and also changed the green color (fourth tile in the third row) to something more visible. I replaced some default colors from the fifth row of the palette with the lighter versions of the Few inspired colors, so I moved these replaced colors to the top row, where the default colors are all too dark to distinguish (though I may decide to bring back the brown color).

– - – – - –

Here is the ColorBrewer2.0 screen for the colors in my earlier palette. Compare the bottom row of the center palette above to the column of colors in the bottom left of the ColorBrewer screen.

And here is an excerpt from Stephen Few’s white paper Practical Rules for Using Color in Charts where he talks about colors. I’ve used Few’s darkest set of colors (middle row) for my dark chart colors, I’ve lightened his medium-dark colors (top row) somewhat for my chart fill colors, and I’ve added highlighting colors which are slightly darker than his light colors (bottom row).

Why am I still worrying about the Excel 2003 color palette? I suppose I should start using Excel 2007 more, especially since the release of the Excel 2010 beta is imminent. But I have not yet become comfortable in Excel 2007, especially for charting, so I would just as soon use 2003. I will keep using 2003 for my chart images, and if a protocol in 2007 differs by enough to confuse readers, I will then capture whatever screen shots are needed.

If you are thinking about colors for a blog or web page theme rather than for Excel, check out the Color Sceme Designer. It is pretty flexible. Its emphasis is on graphic design colors for web page schemes, not for charting (though the two purposes are related).

Update 30 June 2009

I’ve updated the Palette Chooser utility so it includes the original and adjusted Few-inspired palettes. The Palette Chooser was created in Excel 2003, and should also work in Excel 2002 and 2000. I don’t know whether it will work in Excel 97 or in any flavor of MacExcel, and it is irrelevant to Excel 2007.

Download and unzip the Palette Chooser zipped workbook. When the file is opened, the Format menu has an added item called Custom Palette; this menu item is removed when the file is closed.

Colors are stored in the Colors worksheet of PaletteChooser.xls. Palettes are laid out in four-column ranges, with the color index, red, green, and blue values for each color to be modified. Follow the same format to define your own color palette, then assign a worksheet-level name to the range containing your definitions. This name is what is listed in the Palette Chooser dialog.

To modify the palette of a workbook, activate it first before selecting the Custom Palette menu item. The Custom Palette command launches the Palette Chooser dialog. The listbox includes a Reset item, which reverts to the palette which was active in the workbook when the dialog launched. The other listbox items are palettes defined by the named ranges containing the color information.

Here are the Set2_Dark2 (my previous palette) and Paired palettes, based on ColorBrewer:

– - –

Here are the Pastel1_Set1 and Pastel2_Set2 palettes, also based on ColorBrewer:

– - –

Here are the original and adjusted Stephen Few-inspired palettes:

– - –

In the Microsoft newsgroup, someone asked how to hide the data labels in his stacked column chart if the values are zero.

The problem is, a stacked column data point with a zero value has height of zero, and the label sits on the boundary between the two points on either side. In the example below, the labels for series C are fine, but the labels for series B in the Beta stack and for series A in the Alpha stack have no points to label.

The trick is to use the value option for the data labels, rather than the series name option. The series names have been replaced by values, and zeros appear where the unwanted series name labels are in the chart above.

Then apply custom number formats to show only the appropriate labels. In Number Formats in Excel I show how the number format provides formats for positive, negative, and zero values, and for text, with the individual formats separated by semicolons:

<positive>;<negative>;<zero>;<text>
 

Apply the following three number formats to the three sets of value data labels:

"A";;;   "B";;;   "C";;;
 

What these formats do is use the characters in quotes in place of any positive numbers, and use “” (from between the semicolons) for negatives, zeros, and text. The undesired labels are now gone. The labels in the number format strings can be longer than a single character, of course; A, B, and C were easy labels to use for this illustration.

Mike and I have taught advanced Excel classes together, and next month we will host the Second Annual Excel Dashboard Boot Camp. This will be a three-day extravaganza that will take you through the mechanisms of Excel data analysis and charting for purposes of creating effective dashboard-style visualizations.

Mike has written Excel 2007 Dashboards and Reports For Dummies, which is among the most comprehensive Dummies books I’ve come across. The book is geared towards Excel 2007, but its principles and techniques are valid for any version of Excel. Mike takes his readers through dashboard principles, data preparation and chart creation, advanced data techniques including pivot tables and dynamic ranges, automation of dashboards using VBA, and designing interactive dashboards.

Without further ado, Here’s Mike:

A Few of My Favorite Chart Labeling Tricks by Mike Alexander

Often times, a few basic labeling tricks can enhance your chart labels and help bring out an extra layer of analysis. Here are a few of my favorite ways to enhance my chart labels.

Conditionally Color the Y axis

Did you know you can apply a conditional format to your chart labels based on their values? Using custom number-formatting, you can repurpose the Excel’s number formatting to apply your own conditions. For example, if you highlight a range of cells and go to the number format dialog box, you can click on Custom in the ‘Category’ list Then you can enter in something like [Red][<=200]#,##0;[Blue][>200]#,##0.

This will color any number <= 200 Red, and any number >200 Blue.

Charting this range will give you colored chart labels! You can’t do this with conditional formatting.

This will color any number <= 200 Red, and any number >200 Blue.

Jon explains these number format techniques in detail in Number Formats in Microsoft Excel.

Add layers to your X-Axis Chart labels

Have you ever created a pivot chart and got results like this?

You’ll notice that the X-axis has layered/nested labels. That is, month labels and associated years Well, you can achieve this result without a pivot chart by simply including another column of labels in your source data as shown here:

Embed Data Values into your X-Axis Labels

We all know that you can use Data Tables to show the actual data values for your charts. But there are situations where you may want to show the values for only one series in your chart. For example, take a look at this chart:

How did I get the Data Table to show only the values for % Labor cost and not People Count? The answer: I didn’t use the Data Table option. I used Excel’s Alt+Enter trick.

If you’re in a cell and you hold down Alt while you hit enter, Excel will place a carriage return character into the cell, forcing your values to show on two lines.

You can take advantage of this behavior in your data labels. As you can see, the source for the Primary X-Axis label is a combination of the Month name and % Labor Cost (separated by a carriage return).

Here is a sample of the formula used to get this setup. After you enter this formula, click in the space between the quotes and hit Alt+Enter. This will trigger the carriage return.

Note the use of the TEXT function. This is necessary so that the number formatting (in this case, percentages) is retained after you apply the carriage return. See this link for more info on that: http://www.techonthenet.com/excel/formulas/text.php.

Again, this is just a formula trick that concatenates all the values I want to see in my X-Axis label.

I introduced a method for placing two legends into a chart in Legend Entry Tricks in Excel Charts, and a few additional tricks were discussed in the comments of that post. These techniques were not what you might call elegant, requiring non-dynamic graphical objects or superposition of multiple charts.

In this post’s panel chart, the legends are not really official Excel chart legend objects. But if there’s one thing I’ve learned, it’s that I can usually coerce one easy to use chart element to substitute for another which is more difficult to deal with.

The clever reader may recognize the legend keys and legend entries in the panel chart as markers and data labels of an XY series added to the chart. The XY series and labels use the following data. The X values in the first column indicate that the points will be aligned above the 14th and 35th columns of the lower panels. To get these numbers I merely counted columns or spaces between groups of columns, starting with 1 for the half-space at the left edge of the chart. The Y values correspond to the vertical scale of the bottom left panel. These values can be adjusted to fine-tune the positions of the fake legend entries.

Copy the first two columns of this data range, select the unlabeled chart. . .

. . . and use Paste Special to add the data as a new series, with category data in the first column. I had already added a handful of XY series to the chart to produce the four vertical scales in the chart, so my newly added series was automagically added as an XY series. If your series was added the the chart as a column series, right click on it, choose Chart Type, and select an XY type. Then you probably have to format the new XY series to use the primary chart axes.

Run, don’t walk, to Rob Bovey’s Applications Professionals web site, and download his Chart Labeler. Rob calls it the XY Chart Labeler, but it works for any chart type that supports data labels. It’s free, easy to install, and easy to use. Add the labels from the third column of the data range shown above to the points of the XY series.

Finally, format the XY series. Remove the connecting line, and fill each marker with the color of the bars that it represents.

Pretty sneaky! That trick alone is worth the price of membership to this blog.

Summary of Legend Entry Order Rules

Before presenting the rules for legend entry order, a couple of terms need to be defined:

Axis Group – An axis group is the collection of all series plotted on an axis. There is always a primary axis group in a chart, and sometimes a secondary axis group.

Chart Group – A chart group is the collection of all series of a given type within an axis group. For example, there may be a primary line type chart group, or a secondary column type chart group.

When you manually reorder series in a chart, the series available for you to reorder at any given time are all part of a single chart group. You won’t even see other series in the list.

Here are the rules for legend entry order, in vertically listed legends, for charts with horizontal category (X) axes and vertical (Y) axes. The rules are explained and illustrated in the article below.

• Series are listed according to chart type, in this order: Area, Column and Bar, Line, and XY.
• Within each chart type, primary axis series are listed first, followed by secondary axis series.
• Series within each chart group are listed according to plot order.
• If the first chart group listed represents stacked series, series within all chart groups are listed in reverse order.

Effect of Series Type on Order of Legend Entries

Series are listed according to chart type, regardless of their plot order. The order of chart types in the legend is area, then column or bar, then line, and finally XY. This matches the bottom-to-top stacking order of the series in the chart.

Here are two combination charts with the same chart types. The area series is listed first and the line series is listed last, regardless of the plot orders of the series (the number at the beginning of the legend entry label).

Effect of Axis on Order of Legend Entries

Series on the primary axis are listed before series on the secondary axis.

In the left hand chart below, the first two series are on the primary axis and the last two are on the secondary axis. In the right hand chart, this is reversed. In both cases, the primary axis series are listed first in the legend.

This relationship is true as long as all series are of the same chart type (e.g., all are area, column, or line chart series).

Effect of Series Type and Axis on Order of Legend Entries

If the combination chart has series on both primary and secondary axes, the plot order is dictated first by chart type, then within each chart type by axis. A combination of series that share chart type and axis is called a chart group.

The area chart series in the charts below are listed first, whether they are plotted on the primary or secondary axis.

Order of Legend Entries in Stacked Charts

Generally series are listed in the legend in the order they are plotted. However, in vertically stacked charts (line, column, and area), if the legend lists the series vertically, the entries are reversed so they match the stacking order: series one on the botton, series two above series one, etc. In the following examples, the series are plotted in the order one, two, and three.

Here are unstacked and stacked line charts, with normal and reverse legend entry order.

Here are unstacked and stacked column charts.

Here are unstacked and stacked area charts.

In a bar chart, whether clustered or stacked, entries in a vertically aligned legend are listed in the same order as they appear in the chart. The alignment matches the order of the bars in the chart, even if the order of the bars is reversed by plotting an axis in the reverse order.

Effect of Stacking, Axis, and Chart Type on Legend Entry Order

The order of all series within a chart group is reversed if the first chart group listed in the legend is stacked.

In the case where all series types are the same on primary and secondary axes, what matters is the stacking of the primary axis series. Unstacked primary series 1 and 2 in the chart at left are listed first, so no series are listed in reverse order, even though secondary series 3 and 4 are stacked. Primary stacked series 3 and 4 are listed first in the chart at right, so primary axis series 3 and 4 are reversed, as are secondary axis series 1 and 2, even though secondary series 1 and 2 are not stacked.

In the case where there are different chart types on primary and secondary axes, the behavior is dictated by the first chart group listed in the legend. In the chart at left, the secondary area series are stacked, so all chart groups (both the secondary area and primary column series) are listed in reverse order, regardless of whether the columns are stacked. In the chart at right, the secondary area series are unstacked, so no chart groups (neither secondary area nor primary column series) are listed in reverse order, even if the primary columns are stacked.

Unique Horizontal Bar Chart Legend Entry Ordering

Here are a clustered bar chart and a stacked bar chart, with series listed in the order they appear.

Here are a clustered bar chart with the category axis plotted in reverse order, and a stacked bar chart with thevalue axis plotted in reverse order. Again, the series are listed in the order that they appear in the chart.

This reordering of a legend based on a reversal of the order of an axis only appears in a bar chart with a vertically aligned legend.

This article continues the discussion, showing how to use the legend components–legend entries and legend keys–to access more advanced legend functionality.

Legend Entries

Legend entries are treated somewhat independently of each other. You can select a legend entry with two single clicks, one to select the whole legend and the other to select the entry itself.

You can format the text of the legend entry. I’ve bolded the legend entry “two” and to enhance the highlighting of this series I’ve used a thicker line in the chart.

You can selectively remove a legend entry by selecting it and deleting it, while leaving the series in the chart.

In order to restore any legend entries to the legend, you must delete the whole legend, then add the default legend back to the chart, and reapply any custom formatting.

Don’t select the legend key before pressing the Delete key . . .

. . . because you will delete the series itself, not just the legend entry.

However, you can select the legend key then format it, which in turn formats the series in the chart.

This last section does not apply to Excel 2007. You cannot select the legend entry in a chart in Excel 2007. I don’t know why this was changed, because it seemed a nice alternative way to access a series when you otherwise could not select it.

Multiple Legends in a Chart

Sometimes people want to split the legend into different parts. This is not possible. However, like many other visual effects in Excel charts, you can fake it. Here I have two charts, my working chart on the left, and a dummy chart on the right. In the working chart, I have removed the entries I don’t want in one of the legends. In the dummy chart I have removed the entries that are in the working chart’s legend.

I then used a screen capture program (I use SnagIt) to copy the legend from the dummy chart, then I selected the working chart, and pasted this image into the chart, and moved it to where I wanted it.

This is not a dynamic solution: if a label or series format changes, the static image which comprises the second legend may have to be refreshed. This is why I make a separate dummy chart as above. Both charts use the same data and series formats, so I only need to repeat the copy and paste to update my dual legend chart.

Follow up posts in the blog: Order of Legend Entries in Excel Charts, which describes some of the intricacies of legend entry order, and Double Legend in a Single Chart, which shows a dummy series trick to add multiple legends to a single chart.

The best legend is actually no legend at all. A legend by its very nature is somewhat separated from the chart, so a reader must move his eyes back and forth to decode the chart. Placing labels directly on or adjacent to the plotted data is more effective, but it can be more work, and for charts with a lot of data that’s entangled together, data labels may not be practical.

Legend Formatting

The default formatting for the borders in a chart is black. The plot area is also filled with a muddy gray color, but I’ve removed that already.

That’s so many dark lines that I forget I’m supposed to look there for some data. This can be improved by removing the outermost border and using a light gray for the plot area and legend borders.

In fact, that’s still too many visible lines, but I’ll use the light gray in this article so the outlines of the plot and legend are clear.

Without any extraneous borders at all, the data stands out in the chart. Ah, now that’s why we made a graph, to show the data.

The problem with a legend is that it’s somewhat removed from the data. You can move the legend so it’s more integrated into the chart.

You could even tweak the colors in the legend, to make the legend blend in.

Of course, the best legend at all is no legend. Identify the data instead with labels attached to the data series.

The labels are a bit close together, but matching their text colors to the series colors improves the separation between adjacent labels.

Legend Placement

The Excel 2003 Format Legend dialog is shown below, with the Placement tab visible. There are five predefined placements, and the default placement is Right.

Here is a chart with the legend in the default Right position.

This chart has its legend in the awkward Left position. If the value axis were on the opposite side of the chart, this would not be so awkward.

This legend is at the top of the chart.

This legend is at the bottom of the chart. This is probably less effective than the top position, because the horizontal axis comes between the legend and the chart it describes.

This legend is in the least useful corner position. Even Excel knows this is an untenable position. When this position is selected, the chart scrunches up into the opposite corner, and the chart’s text autoshrinks to 5.5 pt. When you change the font to a legible size, like 8 pt, the legend moves to near the right position and the chart itself expands to its original size.

The default placements, at least right and top, are okay. But Excel leaves too much space around the legend and between the legend and the rest of the chart. I start with a default legend position, then reposition and resize the legend to suit my tastes.

Automatic Resizing and Rearranging of the Legend

Let’s return to the the Top position to see a few characteristics of the legend. The legend entries are evenly distributed across the legend.

Suppose the labelsin the legend entries different lengths. All of the entries move to maintain a nice distribution.

If the legend entries are about the same width, but the chart itself cannot accommodate them all, Excel wraps them to fit.

Manually Resizing and Rearranging the Legend

When you select the legend, its edges and corners are highlighted with the familiar handles that allow resizing.

When you begin resizing the legend, the boundaries around the legend entries are displayed.

As this legend is made narrower, the first legend entry wraps to two lines. The other legend entries assume the same dimensions as the largest one, and to accommodate the taller entries, the last one is pushed below the bottom edge of the legend.

When the mouse is released, only five entries are visible.

The legend has to be stretched downward to provide room for the last entry.

Now they all fit, but the spacing doesn’t look very nice. Unless all legend entries have to wrap, it’s probably better to make the legend wide enough to prevent wrapping.

As soon as the legend is being manually resized, all legend entries assume the same dimensions. The other five entries in this legend forget that they had resized themselves to accommodate the longer “one hundred” label.

This legend has been rearranged into a 2 row by three column grid.

This legend has been resized to a 3 row by 2 column grid.

By default, the entries are listed in order, filling rows first, then columns. For example, the entries for one and two are side by side.

If you’d like the entries to be listed in order filling columns then rows, with one above two above three, you need to rearrange the order of series in the chart. Series one is plotted first, four second, two third, five fourth, and so on.

Follow up posts in the blog: Legend Entry Tricks in Excel Charts, which will feature formatting tricks enabled by the individual entries in a chart legend, Order of Legend Entries in Excel Charts, which will describe some of the intricacies of legend entry order, and Double Legend in a Single Chart, which shows a dummy series approach to adding multiple legends in a chart.

Although returns vary for individual goods, overall commodities have gained an average 1% a year over the past five years and an average 7% a year since 1988—even after the latest plunge.

The chart isn’t totally bad, either. It makes a good attempt to show the components of the overall commodities data. I wasn’t happy with the rounded rectangle, I felt it obscured the data as much as included it. I also thought the right to left alignment of the chart was backwards, at least for Western audiences.

I proposed moving the individual commodities to the left part of the graph, and the overall commodities and market index to the right. I proposed lighter fill colors for the individual commodities, and labeling that better separated the individual commodities from the combined commodities. Here is my version of the chart:

But some boss, or editor, or client will want an arrow. So I dropped in an arrow that was as unobtrusive as I could make it.

I think it’s an improvement, do you agree?