–Geoff-Hart.com: Editing, Writing, and Translation —Home —Services —Books —Articles —Resources —Fiction —Contact me —Français |
You are here: Articles --> 2020 --> Preparing graphics for a journal
Vous êtes ici : Essais --> 2020 --> Preparing graphics for a journal
By Geoffrey Hart
Previously published as: Hart, G. 2020. Preparing graphics for a journal. <https://www.worldts.com/english-writing/eigo-ronbun69/index.html>
Most journals still produce a printed version, even if they are increasingly emphasizing their Web site and alternative article formats such as PDF and HTML. As a result, it’s still necessary to prepare graphics (photographs, graphs, maps, and other types of visual information) in ways that make them clear and easy to read in print. Yet most of the graphics I receive from my authors have clarity and legibility problems that are easy to solve. In this article, I’ll provide some suggestions on how to avoid the most common problems. Most of these suggestions will also improve the clarity of graphics in electronic versions of your paper.
First, learn the size of the journal’s printed pages. Graphics should be prepared at a size that will fit conveniently within the space occupied by one or two columns of text so that the journal staff will not have to enlarge or reduce the graphic to fit this space. The author guidelines of most journals define these dimensions. After you have prepared your graphics at this size, print a black and white copy and examine it closely. This will reveal whether the text is legible at that size, whether the lines are clearly visible, and whether patterns and symbols are clear and distinct. If the text is even slightly difficult to read, lines are difficult to see, or patterns and symbols are difficult to distinguish, solve the problem before you send the graphic to the journal: enlarge the text, thicken the lines, and choose different patterns and symbols. Many readers, and particularly older readers like me, have difficulty reading small text and distinguishing very small details. Remember that even if it only takes an additional 30 seconds for a reader to read the text in your graphic, that adds up to a large waste of time for the thousands of readers who will read your paper.
It’s also important to remember that the first—and possibly last—readers who will see your graphics are the journal’s reviewers. If you annoy them by making them work hard to understand your figures, you decrease the likelihood that they will accept your paper without major revision: annoyed reviewers always become more critical.
Why did I suggest printing a copy of each graphic in black and white? This quickly reveals whether readers can easily perceive the contrasts between patterns and colors. A surprising number of readers suffer from some form of colorblindness. For example, up to 8% of men and 0.5% of women, depending on the population, may be unable to distinguish between red and green. Differences that are easily visible in black and white are likely to be visible to people with this form of colorblindness. You can also test for problems related to other forms of colorblindness using the Coblis Color Blindness Simulator.
If you’re submitting your manuscript to an English-language journal, only use English fonts in your graphics. Although the use of Unicode to standardize fonts has greatly reduced the frequency of font problems compared with 35 years ago, when I began using computers, there are still occasional problems with symbols changing when a journal applies their own fonts to a manuscript before publishing it. For example, symbols and text from many Asian fonts that use English characters cannot be simply converted into (for example) Times New Roman without occupying a different amount of space, thereby damaging the figure’s design. In some cases, some symbols may change to the wrong character. This is true of many symbols in standard fonts such as Cambria Math that are commonly used to format equations. If the journal’s author guidelines specify what fonts you should use, follow those guidelines rigorously. If they do not, use Times New Roman for the manuscript text, but use Arial or Verdana for text in graphics. These fonts are supported on most computers, and characters in these fonts rarely change when they are converted to another Unicode font.
Avoid patterns such as ///, |||, #, and |||. These often print unclearly and, particularly if the figure is reduced in size by the journal, they may become indistinguishable. This is particularly true in the small boxes ([]) that are used to display these patterns in the key (legend) of a graph. If your software permits, increase the size of these boxes to ensure that the different patterns are clearly visible and easily distinguishable. Solid shades of a color such as black (i.e., white = 0% black, followed by shades of grey ranging from 10% and 20% (light grey) up to 100% black (the darkest possible color) usually work best, particularly if you’re able to use one or two additional colors at no cost. In general, use the darkest shade (e.g., black) for the smallest symbols or areas, such as the smallest bars in a bar chart. The dark color is most easily visible, and will remain visible even for a small symbol or area, whereas white can disappear. Thus, use white with a border of black for the largest symbols or areas. Avoid yellow, since it often becomes nearly invisible when published. Again, printing a copy of your figure in black and white will let you confirm that the different shades and patterns are easily distinguishable.
The journal’s author guidelines will specify the thickness and positioning of lines. Line thickness is often specified in “points”, with each point representing 1/72 inch. Lines should almost never be thinner than 0.25 points, and a thickness of up to 1 point may be required, particularly if you are designing figures that will be examined on mobile devices like a smartphone or tablet computer. For graphs, move the ticks (– and |) that represent the positions of numbers on the axis so that they are outside of the data area, and are beside the numbers that define the scale. This prevents overlapping between the ticks and parts of the dataset. For simplicity, eliminate intermediate numbers (e.g., an axis with numbers at 0, 1, 2, 3, 4, and 5 should be simplified to numbers at 0 and 5, with horizontal – ticks at each number's position between 0 and 5). Note that in English, we prefer intervals of 5 or 10 for numbers on the axis. Other languages may use different multiples; for example, Chinese tends to use multiples of 4.
I’ve written many articles on effective design of graphics, so if this subject interests you, visit my Web site to see those articles: http://geoff-hart.com/articles/bibliography.html#design
©2004–2024 Geoffrey Hart. All rights reserved.