Preparing Graphs for Papers, Theses and Lab Reports
A correct graph can still be rejected, or look amateur, because of its size, resolution or fonts. This guide covers the practical settings that decide whether a figure prints sharply in a journal: dimensions, vector or raster, DPI, font size, colour and a checklist to run before you submit.
Design the figure at its final size
The most common mistake is making a large figure and shrinking it to fit the page. When a 10-inch-wide graph is scaled to a 3.5-inch column, 10 pt text becomes 3.5 pt and unreadable. Instead decide where the figure will be used first and build it at that width, so a label set to 8 pt is 8 pt on the printed page.
Typical journal layouts have a single column about 3.3 to 3.5 inches (84–89 mm) wide, a one-and-a-half column figure around 5.5 inches, and a full-page width of 6.9 to 7.2 inches (175–183 mm). Theses and reports usually use the text width of the page, often about 6 inches. Always check the author guidelines of the specific journal, because limits do vary.
| Use | Width | Notes |
|---|---|---|
| Single journal column | 3.5 in (89 mm) | Keep to 1–3 series; labels at 7–9 pt |
| 1.5 columns | 5.5 in (140 mm) | Good for multi-series plots |
| Full page width | 7.2 in (183 mm) | Multi-panel figures |
| Thesis or report text | 6 in (152 mm) | Match the page margins |
| Slides | 10 in (254 mm) | Use larger fonts |
Vector or raster: pick the right file format
Graphs made from lines, points and text should be saved as vector files: SVG or PDF. They store shapes and text mathematically, so they stay perfectly sharp at any zoom, produce small files, and can be edited in Inkscape or Illustrator. LaTeX users should prefer PDF, and web pages work well with SVG.
A raster file (PNG) is a grid of pixels and has a fixed resolution. Use it when a journal portal or word processor needs a bitmap. Choose PNG over JPEG for graphs: JPEG compression leaves fuzzy halos around thin lines and text. Avoid saving a screenshot, which has screen resolution and loses the sharpness the figure needs.
How many DPI do you need?
Dots per inch only applies to raster images and is simply pixels divided by printed width. A 3.5-inch figure saved at 300 DPI is 1,050 pixels wide. Most journals ask for 300 DPI for figures with shading or photographs and 600 DPI for line art, and some want 1,000 DPI for pure black-and-white line drawings. Higher is not better beyond what they ask for; it only enlarges the file.
Set the DPI relative to the physical size, not the other way round. If a 3.5-inch figure is exported at 600 DPI you get 2,100 pixels, which is plenty. If you need a bigger bitmap, enlarge the export size rather than stretching a small image, because stretching adds blur.
Fonts, line weights and markers
Use a clean sans-serif such as Arial or Helvetica for figure text, with a final printed size between 7 and 10 pt, and use the same font and size in every figure of a paper. Axis labels should be the same size as, or one point larger than, the tick labels. Lines should be thick enough to survive printing: about 0.75 to 1.5 pt for data lines and 0.5 pt for axes, and markers around 4 to 6 pt.
Write units in parentheses after the quantity, with proper superscripts: “Concentration (mg L⁻¹)”, not “Concentration mg/L” in a different style from figure to figure. In Plotiq you type m^2 for m², H_2O for H₂O, 10^{-3} for 10⁻³ and \mu for μ, so the axis text matches your document.
Colour that survives colour blindness and printing
About 8% of men and 0.5% of women have some form of colour vision deficiency, and many readers print in grayscale. Red and green are the classic pair to avoid. The default Okabe–Ito palette in Plotiq (blue, orange, bluish green, reddish purple, yellow, sky blue) was designed to stay distinguishable for all common types of colour blindness.
For maximum safety do not rely on colour alone. Give each series a different marker shape or line style as well, so the legend still works in black and white. The black-and-white palette switches lines to solid, dashed and dotted styles and bars to white, gray and black fills with outlines for this reason.
A checklist before you submit
Run through these points for every figure: the width matches the journal column; the text is between 7 and 10 pt after sizing; each axis has a title and units; tick labels are legible and not crowded; the legend names every series, or the caption explains them; error bars are defined in the caption; line art is exported as vector or at 600 DPI; the colours are distinguishable in grayscale; the file is not a screenshot or JPEG; and a reader could understand the figure from its caption without reading the text.
If you want to rebuild the figure later with a script, the “Copy Python code” option outputs a matplotlib program that redraws the same chart, so the graph is reproducible and can be adjusted for a different journal in a minute.
Frequently asked questions
- What DPI should I use for a journal figure?
- Most journals want 300 DPI for colour and grayscale figures and 600 DPI or more for line art. Vector files such as PDF or SVG have no DPI and are accepted by most journals for graphs.
- Is SVG or PDF better for LaTeX?
- PDF. LaTeX includes PDF figures directly with includegraphics. SVG needs conversion or an extra package.
- Why does my figure look blurry in Word?
- It is probably a low-resolution PNG scaled up. Export at 300 or 600 DPI at the size it will appear, or insert an SVG or EMF file, which Word keeps sharp.
- How large should the font be?
- Between 7 and 10 pt at the final printed size. Build the figure at that size so you do not rely on scaling.
Keep reading
- Which Chart Should I Use? – A practical decision guide for choosing between line, bar, scatter, histogram, box and pie charts, with the question each one answers and the mistakes to avoid.
- How to Make a Calibration Curve – Step-by-step calibration curve: plot standards, fit a line, check R² and residuals, then calculate an unknown concentration, LOD and LOQ from real numbers.
- Error Bars: SD, SE or Confidence Interval? – What error bars show, how SD, SE and 95% confidence intervals differ, how to calculate each from raw data, and which one to put on your graph.
Written and reviewed by the Plotiq Editorial Team. Spotted a mistake? Tell us and we will correct it. See About Plotiq for how these guides are made.