Florence Nightingale's Rose Diagrams

Florence Nightingale was a competent statistician at a time when the discipline barely had a name. She understood that data presented as tables would be read by administrators, filed, and forgotten. Data presented as pictures was harder to dismiss. Her coxcomb diagrams from 1858 were not decorative supplements to her argument. They were the argument, translated into a form that bureaucrats could act on.

The Crimean War Data

During the Crimean War, Nightingale recorded mortality data for British soldiers by cause of death: preventable diseases (primarily cholera and typhus), wounds sustained in battle, and other causes. The numbers were stark. In the period from April 1854 to March 1855, deaths from preventable disease vastly outnumbered deaths from combat wounds.

This was not a subtle difference. Deaths from disease exceeded deaths from wounds by a factor of roughly seven to one at the peak. The British public and military establishment had assumed the opposite: that war deaths meant battle deaths. The data contradicted this assumption directly. Sanitary reform in military hospitals could, in principle, reduce overall mortality more than any tactical innovation.

Nightingale knew this. The challenge was making decision-makers know it too, and act on it.

The Polar Area Chart

The diagram Nightingale developed is called a polar area chart or coxcomb chart. It resembles a pie chart, but the sectors are not sized by angle. Each sector spans an equal angle (representing one month), and the area of the sector encodes the quantity of interest.

The area of a wedge with radius $r$ and angle $\theta$ (in radians) is:

$$A = \frac{1}{2} r^2 \theta$$

With $\theta$ fixed and equal across all sectors, area is proportional to $r^2$. To double the visual area, you multiply the radius by $\sqrt{2}$, not by two. This means that a sector representing twice the deaths does not look twice as long. It looks twice as large in area, which is the correct visual encoding.

Nightingale divided each month's sector into three colors: blue for preventable disease deaths, red for wound deaths, and black for other causes. The blue region dominated nearly every month. In some months, it swallowed the red region entirely.

Why Not Just a Bar Chart

A bar chart of the same data would have communicated the numbers accurately. Nightingale almost certainly could have made one. She chose the polar area format instead, and the choice was not arbitrary.

The circular structure maps naturally to time: twelve months completing a cycle, the visual return to the starting point emphasizing the persistence of the pattern. The format also allowed the two periods of the war (pre- and post-sanitary reform) to be displayed as two separate diagrams side by side, with the second diagram visibly smaller across nearly every sector. The before-and-after comparison was immediate. No arithmetic was required.

Bar charts require the reader to hold numbers in working memory long enough to compare them. The coxcomb required only that the reader see which color was dominant, which the human visual system can do without effort.

This is not a coincidence. Nightingale was working in a tradition of what we would now call persuasion design. Her audience was parliament and the military hierarchy. They were not going to read tables. A diagram they could look at for 30 seconds and understand was more powerful than a table they would not examine.

Visualization as Rhetoric

The design choices in any visualization carry an argument. The choice of what to put on each axis, what colors to use, what baseline to adopt, what range to show: each of these is a decision, and decisions have effects on what the viewer concludes.

This is sometimes discussed as a problem with visualization, the accusation that charts manipulate. Nightingale's example complicates that framing. Her data was accurate. Her argument was correct. Soldiers were dying unnecessarily from preventable disease, and reform did reduce mortality. The design choices that made the argument legible were also the choices that made it effective.

The rose diagram does what a well-placed example does in a technical article: it makes something abstract accessible to an audience who might not engage with it in its native form. Nightingale did not simplify her data. She translated it. The distinction matters. Simplification removes information. Translation changes the medium while preserving the content.

What It Established

The coxcomb diagrams are often cited as one of the earliest examples of statistical visualization used explicitly for advocacy. What they demonstrate is that the choice of representation is not a neutral act that follows from data collection. It is a decision with consequences, and those consequences can be engineered.

Nightingale's reform campaign was successful. Sanitary conditions in military hospitals improved. Mortality rates dropped. The diagram was not the only cause, but it was a significant instrument. It made the argument visible in a room full of people who controlled the policy levers and had not previously understood what the data said.

The data was available before the diagram. The action followed the diagram. That sequence is worth studying.