The epidemic was measured by subtraction, not by argument

When an epidemic moves through a major population center, the earliest casualty is frequently the accuracy of the official paperwork. Medical professionals rarely have the time to categorize every pathology they encounter. This creates a profound gap between the biological reality unfolding on the streets and the bureaucratic reality preserved in the archives. In the late nineteenth century, the city of Paris provided a perfect demonstration of this exact friction between clinical diagnosis and statistical truth; doctors recorded causes of death based on immediate symptoms, documenting pneumonia, heart failure, and general weakness without ever recognizing the underlying pathogen driving these terminal events. The resulting official archive presented a completely distorted historical picture. To understand the true scale of the event, someone had to stop reading the specific causes of death and start reading the raw volume of mortality.

The numbers paint a picture of sudden devastation that the medical consensus completely missed. From December 15, 1889, through January 31, 1890, the registry in Paris recorded a staggering 12,500 deaths across the city. The logistics of processing that many bodies in weeks must have strained every municipal system to its absolute breaking point. However, the certificates produced during those grueling weeks told a subdued story. According to an essay published by NLM Circulating Now on July 14, 2022, influenza was explicitly indicated in a minuscule fraction of the paperwork. Exactly 243 certificates carried that diagnosis. If a modern researcher were to blindly trust the designated cause of death column, they would conclude that the virus was a minor nuisance rather than a historic catastrophe, missing the invisible tragedy hiding within the aggregate data.

This discrepancy is where the statistician Jacques Bertillon introduced a method that changed how we measure public health crises. In his report for the 1890 Paris statistical yearbook, which saw publication in 1892, Bertillon sidestepped the unreliable clinical diagnoses entirely. He understood one thing clearly. Doctors might mislabel the disease, but they rarely failed to record a corpse. His method relied on establishing a firm historical baseline to serve as a mathematical counterweight to the chaos of the present moment, allowing the normal rhythm of mortality for the metropolis to become a stable foundation for comparative analysis. By analyzing the preceding years, he determined that the three-year average for that precise winter period stood at 7,500 deaths. This baseline accounted for the usual winter ailments, the standard rate of accidents, and the expected natural decline of the elderly population.

The revelation emerges entirely from subtraction. You take the observed reality of the crisis, which is 12,500 deaths, and you subtract the historical expectation of 7,500 deaths. The mathematical remainder is approximately 5,000 excess deaths. These 5,000 individuals represent the invisible toll of the outbreak. Those deaths sit above the ordinary level for that season, and the certificates recorded the immediate mechanism of death rather than the epidemic. The subtraction gives their number. It does not assign a cause to any individual among them, and Bertillon did not claim that it could. This focus on aggregate anomalies rather than individual case files mirrors the analytical shift explored in the doctor who counted deaths, where the narrative emphasizes the power of looking past the obvious clinical labels.

The power of excess mortality calculations lies in their ability to bypass human error and diagnostic limitations by treating every death as an undeniable data point against a historical baseline.

The mechanics of this calculation require a profound discipline in data hygiene. The competent statistician must actively resist the temptation to argue. They cannot debate the individual physician about the accuracy of a specific certificate. Instead the method rests on comparability: the same city, the same weeks of the year, the same registration practice, measured against the seasons immediately before. A rise above that level marks something the ordinary years did not contain. Naming what it was remains a separate argument, made from other evidence. The beauty of the metric is its complete indifference to medical nomenclature. It captures the undiagnosed, the misdiagnosed, and those who died from secondary systemic failures.

While the arithmetic is elegantly simple, the application of this historical data requires strict adherence to several methodological boundaries. The historical findings apply exclusively to Paris during that specific window. We cannot responsibly extrapolate these local urban mortality patterns to other European cities or to the surrounding rural provinces because each geographical area possesses its own unique historical baseline and its own distinct demographic vulnerabilities. Furthermore, the foundational mortality data originates directly from nineteenth-century municipal reports. This makes the raw information well over a century old. This temporal distance demands a level of analytical caution, as the underlying mechanisms of civic record keeping were profoundly different from modern digital registries.

Later academic and biographical literature provides additional context surrounding this era of statistical development, though these sources come with their own distinct limitations. The MacTutor biography serves as a secondary historical source. It specifically highlights the years 1893 and 1900. Similarly, modern researchers have attempted to revisit the Parisian data using contemporary analytical frameworks. A notable example is an article published by Valleron in the journal PNAS in 2010. This modern retrospective introduces the specific figures 408, 2.1, and 1 into the academic conversation regarding the outbreak, but the exact methodology yielding these modern calculations remains obscured because the article is protected behind a publisher paywall. Without transparent access to the underlying computational models, these modern figures must be cited with appropriate caveats regarding their verifiability.

What the method delivered in that winter was a size, not a verdict on any single certificate. Bertillon did not have to persuade a physician to change a diagnosis, and he did not claim to know which household had lost someone to influenza and which had not. He compared a season against the seasons before it and reported the difference. The figure belongs to Paris and to that winter, and the discipline it demonstrates is narrower than it looks: when the record of causes is unreliable, count the totals it cannot distort.