The Definitive Guide to Fonts & Font Sizes in Academic Paper Figures

Ever had your paper bounced back because "figure does not meet formatting requirements"? Or your advisor telling you "the font is wrong and the labels are too small"?

Fonts and sizes seem trivial, but they're one of the most overlooked — and most error-prone — parts of academic publishing. This guide covers the formatting rules for both Chinese and English social science journals, and gives you copy-paste-ready Matplotlib configs.

Why Fonts & Sizes Matter

Your figures will end up in a journal PDF, in print, or on a phone screen. The goal is simple: stay readable at the final rendered size.

Journals shrink your figures to fit single- or double-column widths. If your labels start at 8pt and the figure gets scaled down 50%, you end up with 4pt text — nearly invisible. Too large, and you waste space.

The right approach: set figsize in Matplotlib to the journal's final dimensions, then choose font sizes that work at that size. Don't draw a huge figure and shrink it later.

Wrong vs Correct approach to figure sizing

National Standards & Common Conventions

Chinese academic paper formatting follows GB/T 7713.2—2022 (Rules for Academic Paper Writing), plus university thesis templates and journal submission guides. While the national standard is loose on figure internals, a strong consensus has formed.

Standard Text Elements in Figures

Element	Font	Size	Notes
Figure caption (below figure)	Song / Hei	9–10.5 pt	Centered; "Figure 1" in bold
Axis labels	Song	7.5–9 pt	Variables in italic, units in roman
Tick labels (numbers)	Times New Roman	7.5–9 pt	Always use a Western font for numbers
Legend text	Song + Times New Roman	7.5–9 pt	Mixed CJK + Latin
Figure notes (source, annotation)	Song	9 pt	Below the figure

Key rule: Chinese text in Song (宋体) or Hei (黑体) for bold headings; English and numbers in Times New Roman. This is the universal font pairing in Chinese academic publishing.

Major Chinese Journal Requirements

CSSCI journals (general):

Notes, source lines: Song or Times New Roman, 9 pt
Figure number + caption: centered below figure, 12 pt Song
In-figure text (axes, labels): Song, 7.5 pt
Table text: 9–10 pt, as long as it's legible

**Top journals (e.g. Management World):
Use concise Chinese labels for variables — avoid raw English variable codes

Format regression output cleanly; never paste raw software output

Thesis (general):
Figure notes: Song, 9 pt, 1.25× line spacing, centered

Figure captions go below the figure; table captions go above the table

Chinese Size Names → Points Conversion

Rule of thumb: For Chinese paper figures, 9 pt (小五) is the safest choice. When in doubt, use 9 pt — it won't waste space, and it stays readable after scaling.

English Social Science Papers — Font & Size Standards

APA Style (Most Common in Social Sciences)

APA 7th Edition requirements for figures:

Font: Sans-serif (e.g., Calibri, Arial)

Size range: 8–14 pt

Axis labels: Title Case, parallel to the axis

Legend: Title Case, inside the figure or below it (not on the side)

Avoid: Gridlines, 3D effects (unless truly needed)

Element	Requirement
Font	Arial or Helvetica (sans-serif, mandatory)
Text size	5–7 pt
Subplot labels (a, b, c)	8 pt bold, lowercase roman
Minimum size	5 pt
Single-column width	89 mm (3.5 in)
Double-column width	183 mm (7.2 in)
Max height	247 mm (9.7 in)
Resolution	Min 300 dpi, recommended 450 dpi+
File format	PDF or EPS (vector)

Element	Requirement
Font	Helvetica / Times New Roman / Arial / Cambria
Recommended size	9–10 pt (minimum 6 pt)
Axis labels	8–10 pt
Subplot labels	8 pt Times New Roman, format: (a) (b) (c)
Single-column width	3.5 in (88.9 mm)
Double-column width	7.16 in (182 mm)
Line width	Min 0.5–1.0 pt
Resolution	Line art >600 dpi; grayscale/color >300 dpi

Element	Requirement
Font	Arial / Helvetica / Times New Roman / Courier
Text size	7 pt (at final print size)
Sub/superscript	Min 6 pt
Single-column width	90 mm
1.5-column width	140 mm
Double-column width	190 mm
Resolution	Line art 1000 dpi; mixed 500 dpi; photos 300 dpi

Element	Requirement
Font	Helvetica or Arial (sans-serif)
Size range	8–12 pt (2–3 mm)
Resolution	Line art 1200 dpi; grayscale 300 dpi

Element	Requirement
Body font	Times New Roman
Figure format	Vector PDF / EPS / AI
Photo resolution	300 dpi
Special rule	No asterisks for significance — use standard errors in parentheses

Major English Journal Requirements

Nature / Nature Human Behaviour

Element Requirement
Font Arial or Helvetica (sans-serif, mandatory)
Text size 5–7 pt
Subplot labels (a, b, c) 8 pt bold, lowercase roman
Minimum size 5 pt
Single-column width 89 mm (3.5 in)
Double-column width 183 mm (7.2 in)
Max height 247 mm (9.7 in)
Resolution Min 300 dpi, recommended 450 dpi+
File format PDF or EPS (vector)

IEEE (Engineering & Interdisciplinary)

Element Requirement
Font Helvetica / Times New Roman / Arial / Cambria
Recommended size 9–10 pt (minimum 6 pt)
Axis labels 8–10 pt
Subplot labels 8 pt Times New Roman, format: (a) (b) (c)
Single-column width 3.5 in (88.9 mm)
Double-column width 7.16 in (182 mm)
Line width Min 0.5–1.0 pt
Resolution Line art >600 dpi; grayscale/color >300 dpi

Elsevier Journals (Many SSCI Social Science Titles)

Element Requirement
Font Arial / Helvetica / Times New Roman / Courier
Text size 7 pt (at final print size)
Sub/superscript Min 6 pt
Single-column width 90 mm
1.5-column width 140 mm
Double-column width 190 mm
Resolution Line art 1000 dpi; mixed 500 dpi; photos 300 dpi

Springer Nature Social Science Journals

Element Requirement
Font Helvetica or Arial (sans-serif)
Size range 8–12 pt (2–3 mm)
Resolution Line art 1200 dpi; grayscale 300 dpi

American Economic Review (AER)

Element Requirement
Body font Times New Roman
Figure format Vector PDF / EPS / AI
Photo resolution 300 dpi
Special rule No asterisks for significance — use standard errors in parentheses

Common Patterns Across English Journals

Rule of thumb: For English papers, 8 pt + Arial or Times New Roman is the safest default. When in doubt, check your target journal's Author Guidelines.

Complete Matplotlib Configurations

Chinese Paper Config (Song + Times New Roman)

The core requirement: Chinese in Song (宋体), English and numbers in Times New Roman.

Option A: Global rcParams (Recommended)

python
import matplotlib.pyplot as plt from matplotlib import rcParams config = { "font.family": "serif", "font.serif": ["SimSun"], # Song Ti (Windows) / "Songti SC" (macOS) "font.size": 9, # Global default = 9 pt "mathtext.fontset": "stix", # Math font close to Times New Roman "axes.unicode_minus": False, # Fix minus sign rendering "axes.labelsize": 9, # Axis labels "axes.titlesize": 10.5, # Figure title = 10.5 pt "xtick.labelsize": 9, "ytick.labelsize": 9, "legend.fontsize": 9, "figure.dpi": 300, "savefig.dpi": 600, "figure.figsize": (5.51, 3.54), # Double-column width 140mm } rcParams.update(config)

Usage example:

python
import numpy as np x = np.linspace(0, 10, 100) y1, y2 = np.sin(x), np.cos(x) fig, ax = plt.subplots(figsize=(5.51, 3.54)) ax.plot(x, y1, label="Sine", linewidth=1.0) ax.plot(x, y2, label="Cosine", linewidth=1.0, linestyle="--") ax.set_xlabel("Time (s)") ax.set_ylabel("Amplitude (V)") ax.set_title("Trigonometric Comparison") ax.legend() # Force tick labels to Times New Roman for label in ax.get_xticklabels() + ax.get_yticklabels(): label.set_fontname("Times New Roman") plt.tight_layout() plt.savefig("figure_cn.pdf", bbox_inches="tight") plt.show()

Option B: The mksci-font Library (Simplest)

python
from mksci_font import config_font, mksci_font # Global config config_font({"font.size": 9}) # Or use decorator class="syn-decorator">@mksci_font(ylabel="Y-Axis Label") def my_plot(): _, ax = plt.subplots(figsize=(5.51, 3.54)) ax.plot([1, 2, 3], [4, 5, 6]) return ax my_plot()

Option C: Matplotlib 3.6+ Font Fallback

Matplotlib 3.6 introduced font fallback — list multiple fonts and it picks the right one per character:

python
import matplotlib.pyplot as plt plt.rcParams["font.family"] = ["Times New Roman", "SimSun"] plt.rcParams["axes.unicode_minus"] = False

English/numbers render in Times New Roman; Chinese characters automatically fall back to Song. This is the simplest approach today.

English Paper Configs

General Academic (Serif / Times New Roman)

python
import matplotlib.pyplot as plt plt.rcParams.update({ "font.family": "serif", "font.serif": ["Times New Roman"], "font.size": 8, # 8 pt — the universal safe choice "axes.labelsize": 8, "axes.titlesize": 9, "xtick.labelsize": 8, "ytick.labelsize": 8, "legend.fontsize": 7, "xtick.direction": "in", # Ticks point inward "ytick.direction": "in", "xtick.minor.visible": True, "ytick.minor.visible": True, "axes.linewidth": 0.5, "lines.linewidth": 1.0, "figure.dpi": 150, "savefig.dpi": 600, "savefig.bbox": "tight", "pdf.fonttype": 42, # Embed as TrueType (best compatibility) "ps.fonttype": 42, })

Nature Style (Sans-Serif / Arial)

python
import matplotlib.pyplot as plt plt.rcParams.update({ "font.family": "sans-serif", "font.sans-serif": ["Arial", "Helvetica", "DejaVu Sans"], "font.size": 7, # Nature standard: 5–7 pt "axes.labelsize": 7, "axes.titlesize": 7, "xtick.labelsize": 6, "ytick.labelsize": 6, "legend.fontsize": 6, "axes.linewidth": 0.5, "lines.linewidth": 0.75, "xtick.major.width": 0.5, "ytick.major.width": 0.5, "xtick.major.size": 3, "ytick.major.size": 3, "figure.dpi": 150, "savefig.dpi": 450, # Nature recommends 450 dpi+ "pdf.fonttype": 42, "ps.fonttype": 42, }) # Nature single-column figure fig, ax = plt.subplots(figsize=(3.5, 2.5)) # 89 mm ≈ 3.5 in

IEEE Style

python
import matplotlib.pyplot as plt plt.rcParams.update({ "font.family": "sans-serif", "font.sans-serif": ["Helvetica", "Arial"], "font.size": 9, # IEEE recommends 9–10 pt "axes.labelsize": 9, "axes.titlesize": 10, "xtick.labelsize": 8, "ytick.labelsize": 8, "legend.fontsize": 8, "axes.linewidth": 0.5, "lines.linewidth": 1.0, "figure.dpi": 150, "savefig.dpi": 600, "pdf.fonttype": 42, }) # IEEE single-column fig, ax = plt.subplots(figsize=(3.5, 2.5)) # 88.9 mm ≈ 3.5 in # IEEE double-column fig, ax = plt.subplots(figsize=(7.16, 4.0)) # 182 mm ≈ 7.16 in

SciencePlots Library (One-Line Journal Styles)

python
import matplotlib.pyplot as plt import scienceplots # General science style plt.style.use("science") # IEEE style plt.style.use(["science", "ieee"]) # Nature style plt.style.use(["science", "nature"]) # Chinese support plt.style.use(["science", "no-latex", "cjk-sc-font"]) # Custom overlay plt.style.use("science") plt.rcParams.update({ "font.family": "serif", "font.serif": ["Times New Roman"], "font.size": 8, })

Quick Reference: figsize by Journal

Best Practices Checklist

Before Plotting

Read your target journal's Author Guidelines / Figure Guidelines

Set figsize to the journal's final dimensions (don't scale after the fact)

Pick the right font family (serif vs. sans-serif, per journal rules)

Set global font sizes via rcParams

While Plotting

Axis labels include variable name and unit: Variable (unit) or Variable / unit

Variables in italic, units in roman

All figures in the same paper use consistent font, size, and line width

Place legend inside the figure or below it — never on the side

Use colorblind-friendly palettes (viridis, Paul Tol, etc.)

When Saving

Prefer vector formats (PDF / EPS)

For bitmaps: min 300 dpi (line art: 600 dpi+)

Use bbox_inches="tight" to avoid clipping

Set pdf.fonttype: 42 and ps.fonttype: 42 to embed fonts as TrueType

Quality Check

Verify text is readable at 100% zoom (no smaller than 5–6 pt)

Confirm fonts are properly embedded in the exported file

Test grayscale printing — make sure meaning doesn't depend solely on color

Complete Example: Coefficient Plot for an Empirical Paper

A ready-to-use Matplotlib example for social science regression results:

python
import matplotlib.pyplot as plt import numpy as np # ========== Global Config ========== plt.rcParams.update({ "font.family": "serif", "font.serif": ["Times New Roman"], "font.size": 8, "axes.labelsize": 9, "axes.titlesize": 10, "xtick.labelsize": 8, "ytick.labelsize": 8, "legend.fontsize": 7, "xtick.direction": "in", "ytick.direction": "in", "axes.linewidth": 0.5, "lines.linewidth": 1.0, "figure.dpi": 150, "savefig.dpi": 600, "pdf.fonttype": 42, "ps.fonttype": 42, }) # ========== Data ========== variables = ["Education", "Experience", "Gender", "Urban", "Industry"] coefficients = [0.082, 0.015, -0.156, 0.043, 0.027] ci_lower = [0.065, 0.008, -0.198, 0.012, -0.005] ci_upper = [0.099, 0.022, -0.114, 0.074, 0.059] errors = [[c - l for c, l in zip(coefficients, ci_lower)], [u - c for u, c in zip(ci_upper, coefficients)]] # ========== Plot ========== fig, ax = plt.subplots(figsize=(3.5, 2.8)) # Single-column width y_pos = np.arange(len(variables)) ax.errorbar(coefficients, y_pos, xerr=errors, fmt="o", color="#2c3e50", markersize=4, capsize=3, capthick=0.8, elinewidth=0.8) ax.axvline(x=0, color="gray", linestyle="--", linewidth=0.5) ax.set_yticks(y_pos) ax.set_yticklabels(variables) ax.set_xlabel("Coefficient Estimate") ax.set_title("OLS Regression Results") # Remove top and right spines ax.spines["top"].set_visible(False) ax.spines["right"].set_visible(False) plt.tight_layout() plt.savefig("regression_coefplot.pdf", bbox_inches="tight") plt.show()

This coefficient plot is increasingly popular in economics and sociology. Reviewers now often prefer graphical coefficient displays over regression tables alone.

FAQ

Q: What font should numbers use in Chinese paper figures? A: Numbers and English text should always use Times New Roman. This is the standard convention in Chinese academic publishing.

Q: Should I think in "pt" or in Chinese size names? A: In Matplotlib, always use pt (points). Key conversions: 五号 = 10.5 pt, 小五 = 9 pt, 六号 = 7.5 pt.

Q: Nature and AER have very different font requirements — what do I do? A: Nature mandates sans-serif (Arial/Helvetica); economics journals typically use serif (Times New Roman). Always follow the target journal's Author Guidelines.

Q: My Matplotlib figures keep using the wrong font. Help? A:

Confirm the font is installed on your system (e.g., SimSun, Arial)

Clear Matplotlib's font cache: rm -rf ~/.cache/matplotlib/*

Restart Python and re-import matplotlib

On Linux, you may need to manually copy font files into Matplotlib's ttf directory

Q: Fonts aren't embedded when I save to PDF? A: Set plt.rcParams["pdf.fonttype"] = 42 and plt.rcParams["ps.fonttype"] = 42 to embed fonts as TrueType.

Summary

Fonts and sizes aren't minor details — they're the first impression** your paper makes on reviewers. Get them right and you'll avoid revision requests, while making your visualizations look clean, professional, and credible.

The Definitive Guide to Fonts & Font Sizes in Academic Paper Figures

Why Fonts & Sizes Matter

National Standards & Common Conventions

Standard Text Elements in Figures

Major Chinese Journal Requirements

Chinese Size Names → Points Conversion

Major English Journal Requirements

Nature / Nature Human Behaviour

IEEE (Engineering & Interdisciplinary)

American Economic Review (AER)

Common Patterns Across English Journals

Complete Matplotlib Configurations

Chinese Paper Config (Song + Times New Roman)

Option A: Global `rcParams` (Recommended)

Option B: The `mksci-font` Library (Simplest)

Option C: Matplotlib 3.6+ Font Fallback

English Paper Configs

General Academic (Serif / Times New Roman)

Nature Style (Sans-Serif / Arial)

IEEE Style

SciencePlots Library (One-Line Journal Styles)

Quick Reference: figsize by Journal

Best Practices Checklist

Before Plotting

While Plotting

When Saving

Quality Check

Complete Example: Coefficient Plot for an Empirical Paper

FAQ

Summary

The Definitive Guide to Fonts & Font Sizes in Academic Paper Figures

Why Fonts & Sizes Matter

Chinese Social Science Papers — Font & Size Standards

National Standards & Common Conventions

Standard Text Elements in Figures

Major Chinese Journal Requirements

Chinese Size Names → Points Conversion

English Social Science Papers — Font & Size Standards

APA Style (Most Common in Social Sciences)

Major English Journal Requirements

Nature / Nature Human Behaviour

IEEE (Engineering & Interdisciplinary)

Elsevier Journals (Many SSCI Social Science Titles)

Springer Nature Social Science Journals

American Economic Review (AER)

Common Patterns Across English Journals

Complete Matplotlib Configurations

Chinese Paper Config (Song + Times New Roman)

Option A: Global rcParams (Recommended)

Option B: The mksci-font Library (Simplest)

Option C: Matplotlib 3.6+ Font Fallback

English Paper Configs

General Academic (Serif / Times New Roman)

Nature Style (Sans-Serif / Arial)

IEEE Style

SciencePlots Library (One-Line Journal Styles)

Quick Reference: figsize by Journal

Best Practices Checklist

Before Plotting

While Plotting

When Saving

Quality Check

Complete Example: Coefficient Plot for an Empirical Paper

FAQ

Summary

Option A: Global `rcParams` (Recommended)

Option B: The `mksci-font` Library (Simplest)