Counting María: 64, 4,645, 2,975

There is no single honest death count for Hurricane María. There are three layers, and they do not measure the same thing. First: the official certified direct deaths were 64. That is the civil-registry count of people whose deaths were attributed directly to the storm in the old Weather Bureau style, the style that still worked for Georges. It is real. It is also too small for a grid collapse that lasted months.

The Harvard and NEJM household survey is 4,645. The George Washington University figure is 2,975. Do not attach the survey team to the other study's total.

Second: Kishore and colleagues, in the New England Journal of Medicine, May 29, 2018. A household survey estimated 4,645 excess deaths, with a 95 percent confidence interval of 793 to 8,498, for the window of September 20 to December 31, 2017. Excess means deaths above a baseline, not only drowning and trauma. The interval is wide because the sample is a survey, not a census of death certificates.

Third: the George Washington University Milken Institute study of 2018. It estimated 2,975 excess deaths, with a confidence interval of 2,658 to 3,290, from September 2017 to February 2018. That estimate was later adopted as the official figure of the Free Associated State. It is a different window and a different method from Kishore's. It is not a correction of 64. It is a different question.

A compressed headline that gives the NEJM team the George Washington University total mixes the survey that published 4,645 with the number that belongs to the other study. Do not print that compression. If you need a short line, print the three layers and their windows. If you need an adopted official excess figure, say so, and still leave the 64 visible as the direct certified count.

Kishore's window: September 20 to December 31, 2017. GWU's window: September 2017 to February 2018. They overlap. They are not identical.

Counting is a political act only after the methods are stated. This essay states them and then stops. Yabucoa, the millibars, and the Category labels stay in the landfall record. The migration to Florida stays in its own. The grid stays in its own. The dead deserve a page that does not fold them into a wind speed.

Sources: Kishore et al., Mortality in Puerto Rico after Hurricane Maria, New England Journal of Medicine (2018). Freeze 2026-08-22.