Figure captions need not be half a page

Before leaving high school, every scientist should have learned all the things a graph should contain: a descriptive title, labels for every axis, appropriately spaced tick marks, and a legend if necessary. All pretty straightforward, so you would think any figure published in a scientific journal would adhere to this as a minimum. But I’ve come across far too many figures breaking one or more of these rules. The problem is not that people are excluding the information, rather they are putting everything in the figure caption. Consequently, the figure caption ends up being long-winded, procedural, and not at all interesting. Fortunately, it is easy to make the caption succinct and descriptive with a few quick adjustments to the figure.

Add a legend

The simplest problem to overcome is adding a legend—there’s not really much else to say. Just please don’t do something like this:

Without a legend, we have to search through the caption to match the line with its physical quantity. If we add a legend, the figure caption becomes succinct. And, more importantly, we can spend our time doing science (i.e., interpreting the graph’s meaning) instead of wasting it deciphering what each line represents.

There are no set rules for where to put a legend (it needn’t even go inside the axis), so there are no excuses for leaving it out.

Label your colourmap

Next, lets consider having a label on every axis. Any plot without x and y axes labelled hopefully wouldn’t make it through peer review. However, there is often a third axis that gets overlooked: the colour axis. For whatever reason, the colourbar gets separated from its label and ends up in the caption. I’m now forced to read the entire caption just to determine what the colourmap is displaying and what units are being used. Consider the following example:

This doesn’t seem too bad… it’s only four sentences, right? Well, it really should be only one. In any other bit of writing, if you took four times longer to get across your point, it would be considered bad writing. Why should a figure caption be any different? To show you what we should have, here’s an improved version:

Here the figure is self-explanatory. Primarily, the caption simply adds supporting information: the name of the location and time data were recorded.

Remember, when using colour, pay attention to your choice of colourmap.

Do not repeat the obvious

A caption of the form x versus y is bad practice. In any figure, x and y (whatever they are) should be immediately obvious from the axis labels. There’s no need to repeat them in the caption. Yet that is often the case, especially in multi-panel plots where people simply copy verbatim the information already shown on the y axis or the colourbar. For an example, look to the previous section and note how in the improved figure, I don’t even mention panels (a) and (b) individually. Rather, the caption describes the figure as a whole. Think about it… the caption for a multi-panel figure is much more meaningful and interesting if it describes the overarching idea instead of each panel individually.

Author: Ken Hughes

Post-doctoral research scientist in physical oceanography View all posts by Ken Hughes