A number a committee can file away, a shape it cannot

A column of mortality rates can sit in a bound report for months without ever becoming an argument, because a busy official can turn the page, concede that the arithmetic looks sound, and still leave the room unconvinced that anything in his charge has to change. The reports that Florence Nightingale and William Farr prepared in 1858 and then in 1859 were written for exactly that kind of room: a parliamentary audience that already possessed tables, already possessed rates calculated per 1,000 people, and had grown practised at treating those pages as a formality rather than as a claim on the army it administered. The deaths had already been counted. The problem was that the count, once reduced to a respectable rate, had become a page a political room could file unread and still call itself informed.

The table a room can file

Mortality calculated per 1,000 people is a quiet unit. It is also a careful unit, because it lets a reader compare months of different strength without being misled by a raw headcount, and it is the unit those tables used. A compiled dataset of the Nightingale reports records the figures that way, as rates per 1,000, which is the form in which the statistician and his collaborator had already put the deaths before they built the diagram that was meant to outrun the table. A careful unit remains a quiet one in a political room. A rate can be acknowledged, even praised, without anyone having to picture the volume of preventable loss that the rate describes, which is why a room that does not wish to see a problem can live with a table for a long time and treat the delay as diligence. The friction is not between a true number and a false one. It is between a number that can be filed and a form that cannot.

Two reports, not a memoir

The work of 1858 and of 1859 was not a memoir of a hospital ward and not a portrait of a nurse. It was a joint construction, with Farr as the statistician in the partnership, of a graphic that could travel where a column would not: into the hands of people who had no intention of sitting with a rate, and who would still have to administer an army after they had closed the report. The instrument had to work for a reader who would not do the arithmetic a second time. It had to make preventable loss visible as a field of paper rather than as a figure in a cell. The counts belong to the middle of the nineteenth century, now sit in a compiled dataset that aggregates those reports, and are retold on a museum page that does not disclose the original method.

How a polar diagram is built

Each month becomes a wedge. The radius of that wedge is not the quantity the eye is asked to trust. The area of the sector is, because a line can be glanced past and a field of color cannot, and the whole point of the format is to put the quantity where a hurried official still has to see it. Three causes share that area: preventable loss, wounds, and the remainder gathered as other factors, each painted so that a reader can tell which cause owns the paper without descending into the column of rates per 1,000 that the same reports already contained. Preventable loss is the field that dominates, which is the first half of the argument the diagram was built to carry into parliament: in the military events those reports describe, the deaths that could have been stopped occupy more paper than the deaths that came from wounds. The second half of the argument is not a color. It is a cut.

The table held a rate a political room had already learned to file. The diagram held an area of color that the same room could not file unread.

The cut at March 1855

The diagram is not one wheel. It is a pair, split before and after March 1855, so that the same three causes, the same unit of area, and the same military population can be seen on either side of a single hinge. The split marks where the comparison is made, not what caused the change. Whatever a table of rates per 1,000 might have said about a gradual improvement, the two panels put a before and an after on the same desk, and the after occupies visibly less paper than the before. A room that will not read a column still has to notice that the painted field has shrunk. The split does not add a new death to the ledger. It rearranges deaths the ledger already held, so that the difference across March 1855 arrives as a change in area rather than as another row a clerk can initial and pass along. That is the craft: not a better count, but a count placed where inattention has to fail.

What the area was allowed to say

The tables existed before the diagram and were already in official hands. What changed was where the reader had to look, and how long the largest quantity stayed in view. That is a claim about one document aimed at one room in one decade, and it says nothing about whether pictures beat tables in general. A ledger can be exact and still lose an argument it never got to make, as in the case of a hospital that counted deaths.