Every Thursday in seventeenth-century London a sheet was printed that told its buyers how many people had been buried in the city that week and of what they had died. The figures came from women who went to look at the dead. In the middle of the nineteenth century, material prepared for a Royal Commission on the sanitary condition of the army was printed with tables and coloured diagrams, to show that soldiers in their barracks at home died faster than civilians of the same ages, and that the army in the war with Russia had been nearly destroyed by disease rather than by the enemy. Between the two lies much of the history of what it means to argue about health with numbers.
This article asks a narrow question of two documents from that history. Before a count of deaths could persuade anyone with power to act, what had to be decided about the count itself? A tally of burials is not yet an argument. It becomes one when someone settles what is being counted as a death from a given cause, which deaths belong in the total, and against what other number the total is to be set. Those decisions are the definitions and the denominators. The article reads John Graunt's Natural and Political Observations of 1662 and the 1858 reprint on the mortality of the British army for the places where their authors make those decisions, and for what the decisions are meant to do.
The question belongs to a debate in the history of statistics about where the power of a number comes from. On one side, Theodore Porter treats quantification as a technology of distance, pursued most by those whose standing is weak, and Andrea Rusnock ties the trust won by vital accounts to how the numbers were gathered and who vouched for them (Porter 1995; Rusnock 2002). On the other, Alain Desrosières treats statistical objects as conventions of equivalence made alongside states, and James Scott treats state categories as simplifications that override local knowledge (Desrosières 1998; Scott 1998). Ian Hacking, between them, ties the categories of counting to what came to be counted (Hacking 1990). If Porter is right, the definitions in a mortality count are there to let strangers check it; if Desrosières and Scott are right, they are acts of classification by which the dead become an object a government can act on.
The contribution of this article is a test at the scale of the document. Here two texts are read sentence by sentence for the moment at which a definition or a denominator is fixed, who fixes it, and to whom the result is addressed. Beside each text stands a modern summary of the case written by Skeptron, a studio of the RISCON family that publishes questions about the past drawn from public records. The summaries matter for a different reason: they are written now, by a reader trained to ask what a count leaves unsettled, and they show which of the old decisions a careful modern reader still thinks must be kept apart.
The article finds that both works settle their definitions openly and inside the text, and that both are written to persuade named or implied holders of power. Graunt accepts the searchers' coarse categories as sufficient for purposes he states, distrusts them exactly where they matter most, and corrects the count of plague deaths by setting it against ordinary years. The 1858 report fixes the ages of its comparison, chooses which civilians to compare soldiers with, and corrects for the soldiers who left the army to die as civilians. Neither text fits cleanly on one side of the debate. The article does not decide which side is right; it shows what the two documents can and cannot carry for each.
Section 2 lays out the debate and the histories of the two cases. Section 3 describes the texts, the platform records and the study from which this article is drawn. Sections 4 and 5 read Graunt and the 1858 report in turn. Section 6 brings them together, and Section 7 states the contribution and its limits. The placement of the secondary works in the debate, and their application to the two cases, are this article's own; each work is credited only with what it argues.