All 4 books by Edward Tufte now in
paperback editions, $100 for all 4
Visual Display of Quantitative Information
Envisioning Information
Visual Explanations
Beautiful Evidence
Paper/printing = original clothbound books.
Only available through ET's Graphics Press:
catalog + shopping cart
Edward Tufte e-books
Immediate download to any computer
connected to the internet:
La représentation de l'information
quantitative 200 pages $12
La Representación Visual de Información
Cuantitativa 200 páginas $12
Visual and Statistical Thinking $2
The Cognitive Style of Powerpoint $2
Seeing Around + Feynman Diagrams $2
Data Analysis for Politics and Policy $2
catalog + shopping cart
Edward Tufte one-day course,
Presenting Data and Information
San Francisco, February 9, 10
San Jose, February 13
Arlington, March 31
Washington, April 1, 2
Florence Nightingale's statistical graphics

I'm wondering if you've come across this nifty example of a polar diagram invented by Florence Nightingale.

[Editor's note: A good account of Florence Nightingale's statistical diagrams is here, at the website of the Florence Nightingale Museum.]

-- Leska Fore (email)

Response to florence nightingale piechart

I am wondering it the pie chart format is the best way to represent this data. Although the charts are difficult to read, from the text, the data is represented by "areas" measured from the center of the "pie," which implies some linear relationship.

If the length of the line from the center to the edge of each slice (or colored area in each slice) represents the data, then the size of the areas of each slice tend to mask that measure. If, on the other hand, the areas are meant to represent the data, then it is difficult to compare differences in the areas - is one twice as big as another, or only 1.5 times as big, for example.

I would have thought that a stacked bar chart on a time scale would have been a better choice.

Jim Heimer

-- Jim Heimer (email)

Response to florence nightingale piechart

Please don't call them pie charts! :-)

I read a book about a year ago dedicating a chapter to Nightingale Rose charts with pictures of her work plus the designs of others--but I've forgotten the title! If I find it I will post it. But in the meantime since they aren't used much (at least available on the web) you are stuck with an example from my own (amateur) work.

Basically this author points out 3 benefits of Nightingale Rose charts over pie charts: 1) Each slice takes an equal sector of the circle, making labeling much cleaner; 2) Each slice still maintains an accurate area comparison w/ other slices (by making the radius of the slice equal to the square root of the value); and 3) Since each slice is in a fixed position and the charts can be made much smaller than pie charts, small multiples are very easy to create. The example in the forgotton book was a four-sector NR chart with only two datapoints, but the union of each was plotted (something like the number of respondents who were happy or sad, as well as on or off a medication).

Here's my basic small multiple example:

-- Sean Owens (email)

Response to florence nightingale piechart

The book in question is Cleveland's "Elements of Graphing Data." There are several examples in his book.

-- Sean Owens (email)

Response to florence nightingale piechart

In 1963, after 25 years work and deliberation, the Oxford University Press published "The Atlas of Britain", intended as ". . . a cross-section through the middle of the twentieth century . . . an historical document and not a stop-press commentary". The occasional copy turns up second-hand on Amazon and is a collector's item for anyone interested in cartography. It is thoughtfully conceived and beautifully produced: one of the novel features for the time is a transparent plastic overlay that gives additional administrative features that would otherwise clutter the maps themselves.

The atlas uses a variety of graphic techniques, including polar diagrams/nightingale roses. I attach an extract and the key.

In this particular instance I think the map is an heroic failure. There is just too much information to be assimilated at a glance and the overlapping is a problem. Nevertheless, detailed study does reveal patterns; the incidence of quarrying/mining for example in an approximate crescent north of Birmingham, and the difference in investment in new factory space in electrical engineering between Leicester and Rugby.

-- Martin Ternouth (email)

Response to florence nightingale piechart

A Nightingale Rose by any other name would smell as foul?? These Nightingale Roses are also called "sector graphs", but to me they are just a type of pie chart and contain all the disadvantages of pie charts.

What's wrong with pie charts? ET says: "The only worst design than a pie chart is several of them, for then the viewer is asked to compare quantities in spatial disarray both within and between pieces." (Page 178 VDOQI)

ET, I believe, is saying that most of us think linearly while pie charts present information in a circle. They present data in a way that is unnecessarily difficult to read and understand.

The best graphs are those that communicate with ease, are simple, succinct and make a specific point. They are not a puzzle for the reader but a clear statement. The data in her pie charts is better presented in a table. Using, reading and comparing the data would be easier with a table.

Florence Nightingale made an important and lasting contribution to data presentation but things have moved on. By modern standards her graphs are clumsy and over elaborate. Similarly, perhaps, Victorian novels are now sometimes seen as clumsy and over elaborate, yet nonetheless vital in the development of literature.

-- Sally Bigwood (email)

Response to florence nightingale piechart

An account of the Nightingale diagrams is found in Howard Wainer, Visual Revelations: Graphical Tales of Fate and Deception from Napoleon Bonaparte to Ross Perot. Her letter mentioning the diagrams is at http://www.york.ac.uk/depts/maths/histstat/small.htm A google search turns up additional helpful information.

The diagrams improve on pie charts in exchange for a greater difficulty of reading. As is the case for pie charts, the inherent problem is the difficulty of making good comparisons across the wedges, as Sally Bigwood points out. In general, for such small data sets, tables (or for maps, numbers distributed on the surface perhaps sized as names of cities are sometimes sized in proportion to their population) will outperform graphics.

-- Edward Tufte

The best charts ever

The current edition of The Economist, December 22, 207 - January 4, 2008, features three charts that are described as being among history's best. They include the Minard chart of Napolean's Russian campaign, and it is noted that ET 'calls it the best statstical graphic ever drawn', a chart by Florence Nightingale depicting causes of mortality among British troops in the Crimean War that has been discussed on the Ask ET board, and thirdly a chart by William Playfair depicting the price of wheat in Britain, the weekly wages of 'a good mechanic', and the reigns of monarchs.

In order to give the Minard chart its due, the story in the print edition is on three pages, with the Minard chart extending across an extended fold out page. Nice work by The Economist!

The story is available to non-subscribers on the internet: http://www.economist.com/world/europe/displaystory.cfm?story_id=10278643

The story on the internet contains a sidebar with links to related information (including www.edwardtufte.com) that is not found in the print edition.

The brief commentary on the role of economist and statisticians is particularly interesting. It is written that Nightingale was told by William Farr, the Compiler of Abstracts in the General Registry Office, "'Statistics should be the dryest of all reading.' Fortunately she ignored him."

Regards, Ron Hekier

-- Ron Hekier (email)

Is it wrong that I am enraged by a graph?

The graph conceals parts of the data, misrepresents quantity by area, and what is worst of all, *requires explanation*.

As a graphic designer, I often see work by those who aim for visual impact over legibility. Representing raw data visually should reveal, not conceal. If the result is obtuse, either go back to text or use a standard graph format.

Creating a fluid flow of information; making data a joy to read; bringing clarity through the best medium. This is design. It is not drawing pretty wedges and shoehorning the facts in later. The Economist was wrong to list this alongside Minard's.

-- LL (email)

Science News has sparked renewed interest in Nightingale's presentations of the statistics behind Crimean War mortality rates and the use of the coxcomb for these purposes, in their 6 Dec 2008 issue. A sensible defense can be built around the use of this device, vs. using standard charts, as it conveys an easier to grasp sense of the time dimension, which is often lost in a chart.

Julie Rehmeyer's column, "Florence Nightingale: The passionate statistician", can be found here: http://www.sciencenews.org/view/generic/id/38937/title/Math_Trek__Florence_Nightingale_The_passionate_statistician

-- Juan Courcoul (email)

Why Florence Nightingale used the 'coxcomb' format

Anyone who is still wondering what made Florence Nightingale choose the 'coxcomb' format can see the answer in my paper 'Florence Nightingale's Hockey Stick' presented at the Royal Statistical Society in London in 2010. See also a post describing a recent BBC programme on the subject:



Regards, Hugh Small

-- Hugh Small (email)

Threads relevant to design case studies:

Threads relevant to statistical graphics:

Sports data (along with financial and medical data) are an obvious and natural application of sparklines.