NI Lab All articles
Science Communication

When a Figure Leaves the Paper: A Caption Durability Review for Lab Authors

NI Lab
When a Figure Leaves the Paper: A Caption Durability Review for Lab Authors

Photo by Photo by Toon Lambrechts on Unsplash on Unsplash

A figure rarely travels with its paper. It is cropped for a slide, captured in a screenshot, pasted into a group chat, or lifted into a grant update. In each setting, the caption becomes the entire interpretive frame. If that caption was written as a subordinate note to the main text, the figure may arrive stripped of the information a viewer needs to read it responsibly.

At NI Lab, we treat captions as public-facing metadata rather than decorative labels. The goal is not to make every caption exhaustive. The goal is to make it durable enough that a reasonable viewer can understand what the figure shows, what it does not show, and where it came from. The exercise below is a short review that lab teams can run before submission and that student authors can use when preparing figures for posters, slides, or public summaries.

The Caption Durability Problem

Most captions are drafted inside the paper, where the reader has already encountered the research question, the experimental design, and the conventions of the field. That context is invisible in a screenshot. A viewer may see a bar chart with three conditions and no indication of whether the bars represent technical replicates, biological replicates, or separate experiments. A color scale may be clear in the original PDF but ambiguous after compression. A source line may be omitted because the figure was adapted from another publication.

The durability review asks a simple question: if the figure and its caption were separated from the manuscript, would the caption still do its job? This is not a demand for longer captions in every venue. Journals impose limits, and some figures need panel-specific notes. The exercise is a diagnostic. It reveals which information is essential and which can be moved to a key, a supplement, or a companion note for public use.

A Five-Point Stress Test

The most practical version is a five-sentence draft. Each sentence addresses one category of missing information. The sentences need not appear in this order in the final caption, but the author should be able to write them without consulting the manuscript.

  1. Context. What system, intervention, comparison, or setting does the figure represent? Define abbreviations that a non-specialist or adjacent-field reader would not know. State the comparison explicitly rather than relying on panel labels.
  2. Scale. What are the axes, units, transformations, normalization choices, and color scales? If values are log-transformed, indexed, or divided by a control, say so in the caption rather than leaving the reader to infer it from the methods.
  3. Sample size. What does n mean in this figure? Are the points independent cultures, animals, participants, sites, or technical replicates? If individual values are plotted, note that. If exclusions were applied, the caption should at least point to where they are documented.
  4. Uncertainty. What do the error bars or shaded regions represent? Standard deviation, standard error, confidence intervals, credible intervals, and range are not interchangeable. If a statistical test is displayed, name it and give the relevant comparison.
  5. Source. Where did the data come from, and where can a reader find them? Include a dataset identifier, repository DOI, paper DOI, or adaptation credit when applicable. If the figure is adapted or reproduced, state the license or permission basis.

The five-sentence version is a stress test, not a template to paste into every journal. Its value is that it forces the author to confront the information a standalone viewer would lack. The final caption can then be compressed, with the full version retained for slides, public summaries, or supplementary material.

A Worked Example

Consider a weak caption: 'Figure 2. Enzyme activity across treatments. Activity was higher in treatment B.' The caption names a variable and a direction, but it does not say what the system is, what the comparison means, how activity was measured, how many samples contributed, what the error bars show, or where the data live.

A durable five-sentence draft might read as follows. First, context: 'Figure 2. Enzyme activity in cell-free lysates from E. coli expressing wild-type or variant constructs, measured after 30 min at 37 degrees Celsius.' Second, scale: 'Fluorescence is reported in arbitrary units normalized to the no-template control; the y-axis is linear.' Third, sample size: 'Bars show the mean of n = 6 independent cultures per construct, with individual values plotted.' Fourth, uncertainty: 'Error bars indicate 95% confidence intervals; the difference between wild-type and variant 2 was 1.8 arbitrary units (95% CI 0.4 to 3.2).' Fifth, source: 'Data and analysis code are available at the repository DOI listed in the paper, which corresponds to Figure 2 of the associated publication.'

The revised caption is longer, but it is also portable. A viewer who sees only the screenshot can identify the biological system, the measurement scale, the replication structure, the uncertainty display, and the provenance of the data. The author can then decide which sentences to keep in the journal caption and which to place in a public-facing caption or slide note.

Running the Review in a Lab

The review works best when it is separated from the manuscript. Have the author prepare a figure and caption as a standalone page. Give it to a labmate who has not worked on the project, ideally someone from an adjacent field. Ask that reader to answer five questions: What is being compared? What are the units and scale? How many independent samples are represented? What does the uncertainty display mean? Where is the source? If the reader cannot answer, the caption is not yet durable.

A fifteen-minute lab-meeting slot is sufficient for two or three figures. The author records which questions failed, revises the caption, and returns with a compressed version. The exercise is especially useful before conference season, when figures are most likely to be screenshotted and reused in slides.

A Short Checklist for Student Authors

Captions are not administrative afterthoughts. They are the part of the figure most likely to travel alone. A brief durability review before submission can prevent a screenshot from becoming a misreading, and it gives student authors a concrete way to practice the communication duties that surround publication.

All Articles

Related Articles

The Visitor Question Log: Turning a Lab Tour Into a Diagnostic Instrument

The Visitor Question Log: Turning a Lab Tour Into a Diagnostic Instrument

The Misreading Rehearsal: A Fifteen-Minute Drill Before Your Preprint Goes Public

The Misreading Rehearsal: A Fifteen-Minute Drill Before Your Preprint Goes Public

The Outsider Margin: A Three-Reader Test for Abstracts Before They Travel

The Outsider Margin: A Three-Reader Test for Abstracts Before They Travel