The formatting of Nature, Cell, and Science papers improves their readability

My favourite aspect of a Nature paper is the figure captions. Not the paper’s innovative science. Not the paper’s succinct length. The figure captions. Why? Because the journal’s simple act of bolding the first sentence of a figure caption can force authors to clarify the purpose of the figure. This is one of several seemingly minor formatting issues that ultimately improves a paper’s readibility.

1. Provide high-level figure details first
Figure captions, in my opinion, are the least scrutinized component of a scientific paper. Most jump straight into low-level details such as data sources or specifics of individual panels. Nature, Cell, and Science1 resolve this issue by setting in bold the caption’s first sentence. This elevates its importance and ensures that the author use it to provide a high-level description. Other journals may suggest that authors provide a descriptive first sentence. But without the bolding, it’s not evident that the first sentence is any more important than any other.2 Not that a lack of bold text should stop you from providing a good first sentence.

Here’s where you might expect an example from a Nature paper to demonstrate what I mean. However, I can and will use an example from my own thesis instead because I used the bolded first sentence approach throughout.

Borrowing Nature’s approach to formatting a figure caption for my thesis. A bolded first sentence provides a high-level description of the figure and the subsequent text provides the relevant lower-level details. The specific figure itself is unimportant here. And, yes, this is a caption about a caption: a caption-ception.

2. Write as if your reader will skip your Methods section
Almost as a rule, a Methods section follows an Introduction. Not so in Nature papers, where the Methods are placed at the end of the paper. Why do more journals not follow this sensible approach? I say sensible because, let’s face it, Methods are boring. I read an Introduction and get excited about the problem this paper is going to address and look forward to seeing the Results and the Discussion. My enthusiasm then quickly wanes while traversing all the procedural details. I can’t skip the Methods because the paper will have been written such that this section establishes the relevant context and terminology and thereafter assumes the reader completely understands the experiment and hypothesis.

Placing Methods at the end forces the author to incorporate only the most pertinent details necessary to place their results in context. By contrast, having a Methods section near the beginning lets the author off easy and moves the burden to the reader. For example, the author can now write as if they are saying, hey, remember the location of transect A described in Section 2.1, which was created using profiler B described in Section 2.2, for the time frame C described in Section 2.3, well here’s what it shows. Many readers need and want only the scientific result, which should be distilled from, and largely independent of, these contextual details.

Submitting a paper to a typical journal with your methods in an unexpected place would raise eyebrows. The takeaway message, instead, is to make life easier for your reader: where appropriate, move technical details to an appendix; minimize the need for your reader to recall or retain, across different sections, details that are specific to your experiment like the aforementioned time frames, profilers, or transect locations; and strive to address the ‘why’ not the ‘what’. You’re addressing the ‘what’ if you’re making statements like Figure 1 shows how x changes over time. Instead, use Changes in x are driven by y (Figure 1)Indeed, consider such phrasings for titles and headings:

3. State key outcomes in the title and headers
Short scientific papers are based around a single result and a good title will reflect that. Longer papers may be a collection of results that can be captured by several headings. To borrow from Cell’s information for authors, use titles and headings of the form “Factor X requires Factor Y to function in Process Z” rather than “Analysis of Factors X and Y using Approach Q.” The former format influenced how I titled this post and developed the headings, for example.

“Analysis of” has many variants such “effects of” or “influence on”. As alluded to above, these phrases are implied. It is redundant to state that a scientific experiment is an analysis, or looking for an effect or influence. Therefore, exclude such phrases from poster titles, slide titles, figure captions, headings, or any other short section of text. Forcing yourself to write slide titles or section headings in terms of a result rather than the associated method or topic will help clarify the essence of the specific slide or section.

5. Pick the right font
You don’t get to choose the font when submitting to a journal. You’ve just got to hope the journal makes a good choice. Turns out that Nature has thought carefully about this. The latest iteration of the font they use has been engineered for science with attention to details like increased slanting for italics to better differentiate mathematical symbols and larger sub- and superscripts for better legibility.

This effort is undermined, unfortunately, by their use of 9pt text on their PDFs. An extra 2pts would really help.

6. Use abbreviated phrases rather than acronyms
Nature discourages acronyms in the abstract. In my opinion, among others, this should be extended to the whole paper. Of course, acronyms can’t be flat out banned, but there are many cases where using acronyms is not worth saving a few characters. Often a single word or short phrase can be used in place of an acronym.

Two examples of excluding possible acronyms arise in my own work. In my PhD work, I designated the Canadian Arctic Archipelago as simply the Archipelago rather than the CAA. Or, to refer to diurnal warm layers in my current work, I simply drop ‘diurnal’ after its first usage. ‘Warm layer’ is barely longer than the alternative, DWL, (and actually shorter if counting by syllables) but looks better on the page and is easier to read. More importantly, DWL is meaningless to any reader who skips the section (the Methods perhaps) in which it is defined.

5. And a few more for good measure
Other minor formatting issues that will make life that much easier for your reader:

  • Add page or chapter references when citing a book.
  • Use local time in figures, rather than universal time (UTC). The sun shouldn’t rise at 2300 for example. Reserve UTC for archiving data.
  • Label panels of a multi-panel plot directly, rather than listing each title in the caption.
  • Cite review papers judiciously. Don’t be tempted to make a claim supported by a vague reference to a review paper.
  • Always use a legend rather than including details in the figure caption of the form … parameter X (solid line) and parameter Y (dashed line).


1. Presumably there are other journals that use the bolded first sentence approach. But many more do not.
2. Even with the bolding of the first sentence, some Nature paper authors evidently fail to grasp its importance and simply include a sentence of the form ‘Parameter on the y axis against parameter on the x axis‘. As if we can’t tell that from the axes labels … Though it could be worse. In its instructions for preparing a manuscriptScience deems it necessary to spell out the most basic tenet of scientific graphing: Label graphs on the ordinate [y-axis] and abscissa [x-axis] with the parameter or variable being measured, the units of measure in parentheses, and the scale.

Author: Ken Hughes

Post-doctoral research scientist in physical oceanography

4 thoughts on “The formatting of Nature, Cell, and Science papers improves their readability”

  1. Thanks Ken, This is very applicable to my work. I tend to lurk and learn here, I don’t (yet) write papers (altho’ have a few in mind I’d like to) but do a lot of presentations, tender applications and posters.

Comments are closed.

%d bloggers like this: