The List Is the Elegy

I went to read what I'd built the index for.

A few weeks ago I built a database that lets you look up tornadoes by address. The data is the NOAA Storm Events corpus — every recorded tornado, hurricane, flood, lightning strike, blizzard, since 1950. The thing I built is the search surface. The thing I built it for is the records. Last night, late, I queried for the famous ones and read eight of them cold. Joplin 2011. Moore 2013. Smithville. Rainsville. Tuscaloosa. Beauregard. Mayfield. Palm Sunday 1994 in Goshen United Methodist, Cherokee County, Alabama.

The Goshen narrative ends with this:

Cherokee County - (F02O) (M03O) (F04O) (M05O) (F10O) (M12O) (F24O) (M25O) (F34O) (M34O) (M37O) (M38O) (M39O) (M44O) (F50O) (M54O) (F54O) (M64O) (F72O) (M79O)

F is female, M is male, the two digits are age, the O is a location code from a pre-2007 schema I don't fully know. Twenty entries. F02O is a two-year-old girl. M79O is a 79-year-old man.

You read it once and your eye does the reconstruction without permission. M03O F04O M05O — three children under six. M34O and F34O — possibly a couple at age 34. The 1994 NOAA storm-events database had a punch-card-era data-entry format that compressed each fatality into four characters. Nobody at NOAA in 1993 designed this to read as elegy thirty years later. It just does. The format preserves enough structure — by age, by sex, by sequence — that you reconstruct the church without any of the names you'd want and can't have. The toddlers in the front. The elderly in the pews behind. The middle-aged parents whose ages cluster in the 30s and 40s. An Easter morning service in rural Alabama, encoded in a database format that has long since been replaced.

The format was not designed to do that work. The prose around the format was.

Joplin's narrative includes this paragraph: "An additional indirect fatality occurred due to psychological trauma. Three patients on oxygen may have died from generator failure, though sufficient data as to the cause of death was not available." It includes a 56-year-old "later determined to have died of a heart attack" — originally counted as direct, removed from the count when medical review came back. They include him in the narrative, not in the count. They include a police officer who volunteered from another department and was struck by lightning during response.

Tuscaloosa: "An additional 6 people died (indirect) in the months following the tornado from the stress of the event, not direct injury."

Six deaths from "stress of the event." Counted, attributed, included in the official record. Most institutions would not do that work. It's expensive to gather. It's contestable on every line. It reduces the cleanness of the headline number — the round, citable figure that the news cycle wants. NOAA does it anyway. The narratives are full of this. Joplin includes a city ordinance ("Joplin has a city ordinance prohibiting mobile home parks") because two mobile-home deaths in an EF5 that killed 158 is an absence that needs an explanation; the ordinance is the policy lesson encoded in the prose.

Beauregard 2019 is the most legible piece of the same culture. The narrative names every road in sequence: "Lee County Road 812 ... Road 29 ... Road 11 ... Road 36 ... Road 39 ... Road 51 ... Road 38 ... Road 100 ... Road 166 ... Road 165 ..." A dozen more, then Highway 280, then the Chattahoochee River. The roads aren't there for someone reading from a desk a thousand miles away. They're there for someone in Smiths Station who needs to know whether their cousin on Road 39 was in the path. The narrative is a map for the people who are looking. The spatial precision is doing emotional work and informational work at the same time, and the surveyor knew it would.

This isn't part of the EF rating. The EF taxonomy is a Damage Indicator (DI) and a Degree of Damage (DOD), and a wind-speed estimate, and a path length and width. None of the things I just listed — the indirect deaths, the policy variable, the road-by-road map — are required outputs. The surveyors do them anyway. They are more careful than the form asks for.

I wanted to know whether this was true of NOAA broadly or just of tornadoes. I pulled six non-tornado narratives across modalities — wildfires, flash floods, a heat event. The thesis broke in a clean way.

Yarnell Hill 2013 is the cleanest break. Nineteen Granite Mountain hotshots — young men, some of whom we know by name from the obituaries — walked into a fire that turned. The narrative gives them this:

The wind caused the fire to blow up and it overtook 19 Granite Mountain Hot Shots. They all perished.

Then the prose pivots: Type 2 incident management teams, acreage burned, "16 engines, eight water tenders, two crash-rescue vehicles..." Lahaina 2023, the deadliest US wildfire in over a century, reads like a Coast Guard after-action report — vessel-recovery counts, petroleum-product totals. Camp Fire 2018 is a structure inventory: 18,804 buildings destroyed, 153,336 acres. The 86 dead are a single sentence.

Where the dignity register holds: flash floods. Greenbrier West Virginia 2016, fifteen dead, names "asphyxiation (drowning)" with the same pedantic care Joplin gives "later determined to have died of a heart attack"; one body was found six weeks downstream. Kerr County 2025, the recent Camp Mystic flood — "the owner of the camp, two counselors, and 25 girls ages 8-10 years old," and the man who got his family out the bedroom window but couldn't make it himself. Same culture, water apparatus instead of wind. Tucson 2023's heat narrative is the most interesting transposition: heat has no path to walk, so the dignity-work moves to typology of circumstance — outdoor vs. indoor, AC working vs. broken vs. absent, drug toxicity, prior health issues. Each death decomposed into its conditions. Same care. No path to put it on.

The dividing line is who counts the dead.

The National Weather Service forecast offices survey the events they can directly walk: a tornado path; a flash-flood reach; a heat event integrated through medical examiner data. They are the people on the ground after, with tape measures and clipboards and, in the older records, the punch-card formats. The dignity-work register is their culture, expressed in the events that fall inside their professional territory. Wildfires don't. Wildfires are CalFire and USFS and county sheriffs, and the bodies pass through different paperwork. NWS in those events writes operational synthesis — meteorological context, fire weather indices, resource counts. The dead become statistics because the dead are not theirs to attend to.

This is more interesting than "NOAA does dignity-work." The actual claim is narrower and more defensible: a specific surveyor culture, in a specific class of events, has the discipline of more-careful-than-the-form-asks-for. Goshen UMC's twenty alphanumeric codes and Camp Mystic's named cohort are sibling outputs of that culture, thirty-one years apart.

I should be honest about where I am as I'm writing this.

I am at a desk reading a list of dead toddlers. The surveyors who wrote the prose around the list were standing in the rubble of an Easter service. The reach I am making — F02O is a two-year-old girl — is available because the format preserved enough structure for the reach. The surveyors did the work of attribution. I am only the latecomer who notices what was kept.

The format, on its own, is not elegy. The reach to read it as elegy is what we bring. The surveyors did not write the alphanumeric list to make us reach; the database engineers did not preserve it because they thought it was meaningful. The list is preserved because nobody ever bothered to "modernize" it, and the reach is available because nobody erased it. That's most of what dignity-preservation looks like in long-running institutional records: not deliberate authorship, but the absence of deletion.

The deliberate part — the prose around the list, the road-by-road map, the indirect-death attribution — is the surveyor culture. That's the thing I want to be clear about. The list is the elegy because the format kept enough to make the reach possible. The narrative around the list is the dignity-work, which is a different thing, and is the thing the form does not require.

I did not build the index expecting to find this. I built it for "tornado near me" volume — a market keyword, a long-tail SEO play, a thing to put on the internet. What is on the other side of the search box turns out to be a thirty-year archive of one institutional culture caring more than its forms required, preserved partly by deliberate authorship and partly by benign neglect of the punch-card formats. The address-lookup is the surface; the records are the library. I think I prefer the library to the surface, now that I've sat with both.

The roads. The (measured). The 56-year-old's heart attack three weeks later. The 800-pound safe in the wooded area, found with its door ripped off. The candle factory at DI 23, DOD 7. The map for the cousin on Road 39. F02O.

The list is the elegy because we will let it be. The narrative is the elegy because someone wrote it that way on purpose.