Deaths per 1,000 live births, computed from NHGIS county birth and infant-death counts (NHGIS stopped publishing this broken out by race after 1972 — a real limit on what's public, not a gap in this page). A year is only plotted when both the birth and death counts exist for that race in that county.

Where the non-white line runs meaningfully above the white line here, that's a real, large-enough-sample pattern, not noise — and public health research has a documented answer for why a gap like this can persist even where income and access improve: Arline Geronimus's "weathering hypothesis" (University of Michigan, since the early 1990s) found that chronic stress from racism itself accelerates biological aging in a way that shows up directly in pregnancy risk. Nothing here is softened for comfort.

Worth remembering about every category on this page: none of them were ever fixed. Irish and Italian immigrants were widely excluded from "White" in the 19th and early 20th century; Hispanic identity has been reclassified by the Census more than once; today, how a Dominican American is coded can differ by which box they personally check. These labels are the specific administrative categories used at the time — real enough to shape who got what, but constructed, not fixed biological fact. That instability is itself part of the history, not a flaw in how this page reads it.

Per 10,000 population/live births, NHGIS's own pre-computed rates — no race breakdown available for this span. Read this chart alongside the one above it, not as its continuation: 1972 is where the race-specific data stops, not where this one's line "picks up" a story already in progress.

From the same decennial census data behind the Great Migration/settlement story — shown here beside the mortality trends above so you can see both at once. This is not a claim that one caused the other in this window; the two data eras barely overlap (this stops at 1960, the mortality-by-race data above starts its cleanest run around 1942). It's context, placed where you can actually see it, not a proven line of causation.