Case Study 1: Ed Hawkins's Warming Stripes and the Art of Minimal Time Series

Case Study 1: Ed Hawkins's Warming Stripes and the Art of Minimal Time Series

In May 2018, Ed Hawkins, a climate scientist at the University of Reading, published a single image on Twitter. It showed nothing but a row of vertical stripes, each stripe colored from blue (cool) to red (hot) to represent one year of global temperature from 1850 to 2017. No axes, no numbers, no labels. The image went viral. Within months, it had been reproduced in news articles, printed on clothing, projected onto buildings, and used as the backdrop for television broadcasts. The "warming stripes" became the most iconic climate visualization of the decade, and they did so by removing almost everything a conventional chart has.

The Situation: Climate Data Everyone Ignores

By 2018, climate scientists had been publishing charts of global temperature for decades. The IPCC reports included detailed line charts. Newspapers ran occasional temperature infographics. Scientific papers had rich visualizations with error bars, confidence intervals, and explicit trend lines. The data was unambiguous — global temperatures were rising — and the charts were technically competent.

Yet the charts were not persuading the public. Surveys showed that most people understood climate change was happening but did not find it urgent. Climate scientists were frustrated. The data was clear; the charts were clear; why was the message not landing?

Part of the answer was that conventional climate charts are visually demanding. A line chart of global temperature from 1850 to 2020 has about 170 data points, a y-axis in Celsius or Fahrenheit (typically labeled in tenths of a degree), and a horizontal x-axis with dates. To read it, you have to parse the axis labels, understand what "temperature anomaly" means, interpret the scale, and follow the line's trajectory. For a scientist, this takes seconds. For a general reader glancing at a newspaper infographic, it takes more attention than most readers give.

A second issue was that conventional charts invited debate. Critics of climate science would point to short-term dips in the line and argue that warming had "paused." They would argue about the baseline period. They would question the instruments. The line chart, with its axes and numbers and error bars, gave plenty of hooks for skeptical readers to grab onto.

Ed Hawkins thought there might be a better way. He was a mid-career climate researcher at the University of Reading in the UK, not a designer or communications specialist. But he had been thinking about climate visualization for years, and he had noticed that simpler charts sometimes communicated better than complex ones. He wanted something a reader could understand in one second, without any prior training.

The Design: Stripes, Not Lines

Hawkins's idea was to strip everything away except the color. Instead of a line chart with a y-axis showing temperature, he would show a row of vertical stripes — one stripe per year — with each stripe colored from blue (cool) to red (hot) on a diverging scale centered on the long-term average. The hottest years would be deep red; the coolest would be deep blue. The stripes would be arranged in chronological order from left (1850) to right (2017).

No y-axis. No x-axis. No legend. No title (in the original version). Just a row of colored stripes.

The result was visually simple but informationally rich. The warming trend was immediately obvious: the left side of the image was mostly blue and white, while the right side was dominated by red. A reader could grasp the pattern in about one second — "it's getting hotter, a lot hotter." No numerical interpretation was needed. No axis parsing. No argument about baseline periods.

Hawkins posted the first version on Twitter in May 2018 with the caption "Warming stripes for 1850-2017: each stripe represents the temperature of a single year, ordered from the earliest available data at each location." The image was 1500 pixels wide by 300 tall — an unusual aspect ratio for a scientific chart, but perfect for a social media image and for use as a banner on web pages.

The response was overwhelming. Within days, the image had been retweeted tens of thousands of times. Climate scientists shared it, journalists reproduced it in articles, activists printed it on signs at climate protests. Within weeks, Hawkins had made local versions for individual cities and countries, which were shared further. Within months, the stripes were on clothing, coffee mugs, murals, and building facades.

What Makes the Stripes Work

The warming stripes succeed for several specific design reasons.

Extreme visual minimalism. The image has no chart chrome at all — no axes, no labels, no ticks, no grid, no legend. Every pixel is either a stripe or the space between stripes. The data-to-ink ratio (Chapter 6) is as high as it gets. A reader who knows nothing about charts can still see the pattern because there is nothing else to see.

Color as the only encoding. Color does the entire informational work. The diverging blue-to-red palette is intuitive (humans associate red with hot and blue with cold almost universally). The gradient is smooth enough that year-to-year changes are visible but not overwhelming. The contrast between the blue-dominated left and red-dominated right is stark, and this contrast is the main message.

One observation per stripe. Each stripe is one year. The reader does not have to count or interpolate or read off a line. The simplicity of "each stripe = one year" makes the temporal axis obvious without any labels.

A chronological left-to-right ordering. Western readers read left to right, and time is conventionally represented left-to-right in charts. The reader's eye naturally moves from "the past" (left) to "the present" (right), and the warming pattern unfolds with the reading motion.

No baseline debate. The stripes are colored relative to the long-term mean, but the reader does not see the baseline explicitly — they just see colors. A skeptic cannot argue about the baseline choice without first knowing it was a choice at all. The stripes bypass the usual arguments about reference periods because they do not show them.

Shareable aspect ratio. The 5:1 aspect ratio is perfect for Twitter headers, website banners, and clothing prints. A standard square or 16:9 chart would not have the same visual impact or the same compatibility with common display contexts.

Location-specific versions. Hawkins quickly made versions for individual countries, states, and cities. A Londoner could share the "warming stripes for London." A New Yorker could share "for New York." The local versions multiplied the virality by making the visualization personally relevant.

What the Stripes Do Not Show

The warming stripes are a minimalist visualization, which means they omit almost everything a conventional chart includes. Specifically, they do not show:

Numerical values. You cannot read off the temperature in any specific year from the stripes. The colors give rough ordering but not magnitude.
Uncertainty. There is no error bar, no confidence band, no indication of measurement uncertainty.
Causation. The stripes show the pattern but not the drivers (CO2, solar, volcanic, etc.).
Short-term events. A single cool year in a dominantly red region might be a La Niña; a single hot year might be an El Niño. The stripes do not distinguish these from long-term trend.
Method. The stripes do not explain how the data was collected, homogenized, or averaged across the globe.

For a scientist, these omissions are problematic — you cannot do science with stripes. But the stripes were never meant to do science. They were meant to communicate one specific message ("it is warming, rapidly") to a general audience that was not going to read a scientific paper. For that purpose, the omissions are features, not bugs. A chart with error bars and axis labels and methodological footnotes would not have gone viral. The stripes did because they were simple.

This is the core lesson of the warming stripes: design for the audience and the delivery context. A chart in a scientific paper should include error bars because scientists need them. A chart on a T-shirt should not include error bars because the audience is different. The same underlying data supports radically different visualizations, and neither is objectively better. Each is optimized for a different use.

The Broader Impact

In the years since 2018, the warming stripes have become ubiquitous. They appear:

On BBC, ITV, and other broadcast TV sets as backdrop graphics for climate segments.
On the cover of scientific journals like the Bulletin of the American Meteorological Society.
As banners on the climate section of major news websites.
On clothing sold by climate organizations.
On buildings — projected onto facades during climate protests, painted as murals, printed on flags.
In school textbooks and educational materials.
As annual updates — Hawkins still posts new versions each year with the latest data.

The stripes also spawned many variations: stripes for ocean heat, stripes for CO2, stripes for rainfall, stripes for Arctic sea ice, stripes for specific countries and cities. Hawkins set up a website (showyourstripes.info) where anyone can download a locally-relevant version. The meme has become a format — a "warming stripes" visualization is now understood as a specific style, and climate scientists can produce new versions for new variables with no additional design work.

Other fields have adopted the format. There are "democracy stripes" (showing trends in democratic indicators), "inequality stripes" (showing Gini coefficients over time), and "life expectancy stripes" (showing demographic changes). The stripes format has generalized beyond climate into a broader template for "long-term change visualized through minimal color coding."

Theory Connection: When Less Is More

The warming stripes are a case study in the data-ink ratio principle from Chapter 6 taken to its logical extreme. Tufte's original formulation argued that charts should remove everything that is not data-ink. The stripes remove almost all ink except the data — no axes, no labels, no legend, no title. The remaining ink is the minimum necessary to convey the pattern.

The stripes also illustrate the audience-matching principle from Chapter 8. The specific audience is "general readers who will glance at the image for one second." The specific delivery context is "social media and web banners." For these constraints, the stripes are arguably the optimal design. A more detailed chart would convey more information per viewing, but viewers would not linger long enough to absorb it. The stripes match the viewer's attention budget precisely.

A third theoretical point: the stripes demonstrate the power of color as a primary encoding. Most of this textbook has argued for position as the strongest encoding (scatter plots, line charts, bar charts) and color as a secondary encoding (hue, saturation, to distinguish categories). The stripes invert this: color is the primary encoding, and position (left-right) only provides the temporal ordering. This inversion is appropriate when the message is about magnitude-per-time-point rather than about rate-of-change — when you want the reader to see "warm vs. cool" directly rather than inferring it from a sloping line.

The inversion has limits. Color as a primary encoding works for discrete temporal buckets (one stripe per year) but not for continuous data. It works for a single variable but not for multi-variable comparisons. And it relies on the reader's ability to perceive color accurately, which fails for colorblind viewers (Hawkins's palette uses blue-red, which some colorblind individuals cannot distinguish — an ongoing concern that has led to alternative versions in colorblind-friendly palettes).

For most scientific and business contexts, position-based charts remain the better choice. But the warming stripes are a reminder that the "standard" chart design is not the only option, and in specific communicative contexts, an unconventional choice can outperform every conventional one.

Discussion Questions

On minimalism. The stripes remove axes, labels, and legends — all of the things Chapter 7 argued were essential for self-explanatory charts. How do the stripes get away with this? Is the lesson "labels don't matter" or something more nuanced?
On the virality formula. The stripes went viral in a way that most climate visualizations do not. Is there a repeatable formula for viral visualization, or was this lightning in a bottle?
On scientific rigor. The stripes show the pattern but hide uncertainty, causation, and method. A climate skeptic could reasonably argue that the stripes are propagandistic — they bypass scientific discussion by refusing to engage with it. Is this a legitimate critique?
On audience-matching. The chapter argues that design should match the audience and context. Does this justify using the stripes in every context, or are there cases where they would be inappropriate?
On generalization. The stripes format has been adopted for many variables (CO2, inequality, democracy). Are these effective, or are they riding the coattails of the warming stripes without earning the viral moment?
On your own use. The next time you have long-term data and a general audience, would you consider stripes? What specific conditions would make them right or wrong?

Ed Hawkins's warming stripes are one of the most successful science visualizations of the 21st century. They show that a radically minimalist chart, when matched to its audience and context, can communicate more effectively than any conventional alternative. When you build time series visualizations, remember that the default — line chart with axes and labels — is not the only option. Sometimes the right answer is to remove everything except the data-ink and let the color do the talking. The stripes are the proof of concept.