Exercises: Typography, Annotation, and the Words on Your Chart

These exercises are non-programming. They build the mental habit of writing chart text with the same discipline you would apply to chart design. Do them with a pencil and paper, or with a text editor — the point is the writing, not the rendering.

Part A: Conceptual (6 problems)

A.1 ★☆☆ | Recall

Define the "self-explanatory chart standard" in your own words. What does it mean for a chart to be self-explanatory, and why does the chapter argue that this standard is worth pursuing even though most charts will appear in contexts with surrounding prose?

Guidance The standard says a chart should be readable without any surrounding text. Think about the reasons the chapter gives: charts travel, they get decontextualized, and readers often glance at them without reading the accompanying article. Your answer should connect the standard to the 5-second test and to real-world use cases like social media sharing and slide embedding.

A.2 ★☆☆ | Recall

Name the four categories of chart text introduced in Section 7.1 (or five if you count source attribution as its own category) and describe the specific job of each.

Guidance The four are title, subtitle, axis/tick labels, and annotations — plus source attribution as a fifth. Each has a specific role: the title states what the chart is about (or what the finding is), the subtitle provides essential context, the axis labels name quantities and units, and annotations direct attention to specific features. Source attribution enables trust and verification.

A.3 ★★☆ | Understand

Explain the distinction between a descriptive title and an action title. For each type, describe a context in which it would be the appropriate choice.

Guidance A descriptive title states what the chart is about ("Quarterly Revenue"). An action title states the finding ("Revenue Grew 18% in 2024"). Descriptive titles are appropriate for exploratory charts, reference tables, and academic publications with neutral conventions. Action titles are appropriate for business communication, data journalism, dashboards with stable metrics, and any context where the reader needs to grasp the finding quickly.

A.4 ★★☆ | Understand

The chapter argues that stating a finding in the title is not "editorializing" in any objectionable sense. Reconstruct the argument. What is the ethical line between an action title that is defensible and one that overstates the evidence?

Guidance Connect the argument to Chapter 4's principle that every chart is an editorial. A descriptive title does not make a chart neutral — it only hides the editorial stance. An action title makes the stance explicit. The ethical line is between titles that match the evidence the chart actually shows and titles that add interpretation the chart does not support. "Revenue Grew 18%" is defensible if the chart shows 18% growth; "Revenue Collapsed" is overstatement for the same chart.

A.5 ★★☆ | Analyze

The chapter identifies five typographic principles for data visualization (Section 7.2): single font family, size hierarchy, weight for emphasis, meaningful alignment, whitespace around text. For each principle, explain why it matters for chart legibility, not just what it prescribes.

Guidance For each principle, ask: what goes wrong when the principle is violated? Mixing fonts creates visual noise. Flat hierarchy makes it impossible to tell what is important. Using color for emphasis steals color from the data. Haphazard alignment makes the chart feel unprofessional. Cramped text makes everything harder to read. The "why" is usually about cognitive load and reader effort.

A.6 ★★★ | Evaluate

The chapter says that on-chart source attribution is non-negotiable and should be built into your default workflow. Defend or critique this position. Under what circumstances, if any, might a chart legitimately appear without source attribution?

Guidance The chapter's argument is that attribution is a trust mechanism and that charts without attribution can be shared in ways that strip provenance. Consider the counter-argument: exploratory charts in a personal notebook do not need attribution because they are for the maker's own use. Are there other legitimate exceptions? What about charts whose data source is obvious from context (a dashboard where every chart comes from the same internal system)?

Part B: Applied — Writing Titles and Annotations (10 problems)

B.1 ★★☆ | Apply

Rewrite each of the following descriptive titles as an action title. You may need to invent plausible findings to make the exercise concrete — that is expected. Write the action title as if you had seen the data.

  • (a) "Monthly Active Users"
  • (b) "Vaccination Coverage by Country"
  • (c) "Daily Stock Returns"
  • (d) "Energy Consumption by Sector"
  • (e) "Response Time Distribution"
Guidance For each title, imagine a specific chart and a specific finding. Your action title should state the finding as a declarative claim with specific numbers where possible. Compare your title to the descriptive version and ask: which one tells the reader what to conclude?

B.2 ★★☆ | Apply

You have drafted the following action title for a chart about Meridian Corp's new product line: "Product X Is the Fastest-Growing Launch in Company History at 240% Year-Over-Year Growth." Identify at least two potential problems with this title and suggest a revised version.

Guidance Consider: is "fastest-growing in company history" a defensible claim? Is 240% suspiciously high? Does "fastest-growing launch" imply a comparison group that is clear to the reader? Is the title too long for a single line? A revised version might be shorter, more specific, and more defensible against scrutiny.

B.3 ★★☆ | Apply

Write a one-line subtitle to accompany each of these action titles. The subtitle should provide essential context the title does not — time range, geographic scope, units, caveats.

  • (a) Title: "Global Temperatures Have Risen 1.2 Degrees Since 1900"
  • (b) Title: "Meridian Corp Revenue Grew 18% in 2024"
  • (c) Title: "Most OECD Countries Exceed 85% Vaccination Coverage"
Guidance A subtitle typically adds the time range, the data source, the geographic scope, and any important processing notes. Keep it to one or two lines — the subtitle is for context, not for the full story.

B.4 ★★☆ | Apply

For a chart showing Meridian Corp's stock price from 2020 to 2024, write three annotations that call out specific events or features. Each annotation should be under 15 words. Assume the chart shows a sharp drop in March 2020 (COVID), a steady climb from mid-2020 to late 2021, a drop in early 2022 (rising interest rates), and a recovery through 2024.

Guidance Each annotation should be attached to a specific point or region and should explain the feature in a short phrase. Examples: "March 2020: COVID market crash," "Early 2022: Fed rate hikes," "December 2024: New all-time high: $186.30."

B.5 ★★☆ | Apply

You have a chart with five data series (Product A, Product B, Product C, Product D, Product E) and you are deciding whether to use a legend or direct labels. Sketch your decision process. Under what conditions would you choose direct labels, and under what conditions would you fall back on a legend?

Guidance Direct labels work when the series have clean endpoints, limited overlap, and a manageable number (under ten). They struggle with short or crossing series, very dense charts, and charts where label placement would obscure the data. With five series, direct labels are usually feasible unless the lines cross heavily near the right edge of the chart.

B.6 ★★☆ | Apply

A chart shows vaccination rates for 50 U.S. states over time. The reader is a policy analyst in one specific state (say, California). Describe a design that uses direct labeling for California and the highlight strategy for the other 49 states. What would the reader see?

Guidance The California line is drawn in a bright color (the reader's state gets the emphasis). The other 49 lines are drawn in muted gray, providing context without demanding identification. California gets a direct label at the end of its line. The other states may or may not be labeled at all — usually not, because the reader does not need to identify them individually; they are the background.

B.7 ★★☆ | Apply

Take a default-style axis label like "Revenue" or "Users" and rewrite it with inline units for each of these scenarios:

  • (a) The values are in U.S. dollars, with a range from $50,000 to $5,000,000
  • (b) The values are monthly active users, from 10,000 to 10,000,000
  • (c) The values are temperature anomalies in degrees Celsius, from -0.5 to +1.5
  • (d) The values are percentages, from 0% to 100%
Guidance Include the unit in parentheses after the quantity name. Consider scaling to natural units if the raw range is unwieldy. Examples: "Revenue (USD millions)" if you scale to millions, "Monthly Active Users" with millions suffix on the tick labels, "Temperature Anomaly (°C)," "Vaccination Coverage (%)."

B.8 ★★☆ | Apply

For the same axis in B.7a (revenue from $50,000 to $5,000,000), design the tick labels. How many ticks should the axis have? How should the numbers be formatted? What should the highest and lowest labeled ticks be?

Guidance Consider scaling the axis to millions. Ticks at 0, 1, 2, 3, 4, 5 (millions) with an axis label "Revenue (USD millions)" would give you a readable axis with five major ticks. Thousands separators in the tick labels (50K, 100K, etc.) are another option if you keep the raw dollar scale. The key principle: fewer ticks than the default, human-readable numbers, units in the axis label rather than on every tick.

B.9 ★★★ | Apply

Write a complete set of chart text — title, subtitle, axis labels, three annotations, and source attribution — for a line chart of global atmospheric CO2 concentration from 1960 to 2024. The data source is the Mauna Loa Observatory. CO2 has risen from about 315 ppm in 1960 to about 420 ppm in 2024. The rise has been accelerating. 1970 saw about 325 ppm. 2000 saw about 370 ppm. The annual "sawtooth" pattern reflects northern hemisphere seasonal growth.

Guidance Your title should state the finding: the rise from 315 to 420 ppm or the acceleration. The subtitle should name the dataset and the time range. Axis labels should include units (ppm). Annotations might call out the 315-ppm starting point, the current value, and one or two milestones in between. The source attribution should name Mauna Loa Observatory and any processing notes.

B.10 ★★★ | Create

Design a "chart text style guide" for your own work — a one-page specification of the default fonts, sizes, weights, colors, and alignments you will use for titles, subtitles, axis labels, tick labels, annotations, and source attributions. Be specific: exact point sizes, specific font names, specific colors.

Guidance A good style guide specifies enough detail that a colleague could implement it from the document alone. Consider: body font (sans-serif like Inter or Source Sans Pro), title size (18-20pt), subtitle size (12-14pt), axis label size (11-12pt), tick label size (9-10pt), annotation size (9-11pt), source attribution size (8pt). Include weight (bold/regular) for each level. This becomes the basis for your personal matplotlib rcParams in Part III.

Part C: Synthesis and Design Judgment (6 problems)

C.1 ★★★ | Evaluate

Find a real published chart from a news outlet (NYT, FT, The Economist, Washington Post, Reuters). Analyze its chart text:

  • (a) Is the title a descriptive title or an action title?
  • (b) How many annotations does it have, and what do they call out?
  • (c) Is there a source attribution on the chart itself, or only in the caption?
  • (d) What typographic choices (font, size, weight, color) does the chart make?
  • (e) Does the chart pass the 5-second test for a reader who has never seen it?
Guidance Choose a published chart carefully — you will find that major graphics desks almost always use action titles, limited annotations, on-chart source attribution, and clean typography. Identify each element specifically. This exercise is about training your eye to see the craft in professional work.

C.2 ★★★ | Analyze

The chapter argues that annotations are "the text that does the most work per word." Explain what this means. Why is a well-chosen annotation more valuable than an additional sentence in the chart title or subtitle?

Guidance Annotations are attached to specific features in the data. A title statement applies to the whole chart; an annotation applies to one data point or one region. The annotation's value per word is higher because it carries specific information about a specific location. A five-word annotation on the right data point communicates more than a five-word addition to the title.

C.3 ★★★ | Create

Take a chart you made recently that has only a default title (or no title). Write an action title for it. Then write a subtitle. Then identify at least one feature of the data that deserves an annotation and write the annotation. Time yourself — this exercise should take under five minutes if you know your data.

Guidance The whole point of this exercise is the speed: writing chart text is a discipline, not a creative writing exercise, and with practice it should be fast. If you find yourself taking more than five minutes, either the chart's finding is unclear to you, or you are overthinking the words. Settle on a plausible draft and move on.

C.4 ★★★ | Evaluate

Imagine a chart that has a perfect action title and annotations but weak typography — mismatched fonts, no size hierarchy, random alignment. How effective is the chart? Now imagine the reverse — perfect typography but a vague descriptive title and no annotations. How effective is that chart? Which weakness is more damaging, and why?

Guidance The typographic weakness makes the chart look amateurish and slightly harder to read, but the content still communicates. The content weakness (vague title, no annotations) makes the chart feel professional but uninformative — the reader still does not know what to conclude. The content weakness is more damaging because no amount of polish can compensate for the absence of a clear message. This is one reason to prioritize the action title before worrying about fonts.

C.5 ★★★ | Create

Write a one-sentence rule for yourself that you will apply to every chart you make from this point forward. The rule should capture the most important lesson from this chapter, in a form that you can apply without consulting notes. Example formulation: "Every chart I publish will have a title that states the finding, at least one annotation, and an on-image source attribution."

Guidance The point of this exercise is to commit to a specific practice. Your rule should be short, specific, and enforceable. Avoid vague commitments like "I will write better chart text." Be concrete about what you will do every time.

C.6 ★★★ | Evaluate

Consider the position that action titles should be allowed even in academic publications, replacing the traditional neutral figure captions. Argue for or against this position, identifying at least one specific consideration the chapter does not fully address.

Guidance Academic conventions value neutrality and the separation of presentation from interpretation. Action titles blur this separation. Arguments for: academic charts are often seen by non-specialists who would benefit from a clearer message; the finding is not hidden in the prose but stated where it belongs. Arguments against: academic tradition relies on the author's reading being checked by reviewers; action titles could be seen as overreach. This is a legitimate judgment call.

These exercises build the writing habit that separates default-chart practitioners from thoughtful communicators. The exercises in Part B are the most practical — writing chart text is a skill that improves only with practice. Do at least five of the Part B exercises before moving on to Chapter 8.