When the building itself is the threat
Gas leaks, chemical spills, infrastructure failures, and fires occupy a quieter corner of this archive (165 cases, a fraction of the active-shooter chapter's total) but they are also the genre where institutional alert language is at its clearest. The reason is structural: when the threat is the building, the protective action is unambiguous. Evacuate. Everything else in the message is logistics.
The fires and explosions
The headline cases here include the Texas A&M Aggie Bonfire collapse of November 18, 1999 (twelve students killed when a forty-foot stack of logs collapsed during construction: not a fire, but the case is preserved here for how the campus communicated a mass-casualty emergency in the pre-SMS era), Pace University's evacuation of its lower Manhattan campus on September 11, 2001, blocks from the World Trade Center, the UCLA laboratory fire of December 29, 2008 (research assistant Sheharbano Sangji fatally burned during a transfer of t-butyllithium: the case that drove new chemical-hygiene rules across the UC system), the MIT Building 13 chemical spill of February 2019, and the University of Hawaiʻi at Mānoa laboratory explosion of March 16, 2016, in which a postdoctoral researcher lost an arm. Several of these produced multi-day sequences of follow-up communications and after-action changes that are documented in their case files.
Gas leaks: the genre's clearest writing
If you want to read short, well-composed campus alerts, read the gas-leak cases. The protective action is simple and the language shows it. Rowan University's 2014 message reads, in full, "There is a gas leak at the Edgewood Park Apartments, Building 500. Any students that are still inside the building need to evacuate immediately." Its closure, later the same day: "South Jersey Gas is at Edgewood Park Apts & repairs are in progress. The gas leak is under control. Residents can re-enter the building." Name the building, name the action, name the moment it ended. The shape recurs across this chapter's gas-leak cases in state after state.
The interesting cases in this sub-genre are the ones where the incident refused to end on the first message: leaks that outlasted their alert, evacuations that widened, all-clears that had to carve out exceptions. Those second messages are where the genre's edge cases live, and the case files preserve them alongside the tidy ones.
Weather-borne infrastructure failures
A few cases here are infrastructure failures caused by weather: the Hurricane Sandy closures of October 2012 (the University of Delaware's sequence is preserved here), and the sustained boil-water advisories at Jackson State during the August 2022 Jackson water crisis. The alert language for these is its own mode: it has to communicate that a condition will persist for days, not hours, which is a different rhetorical task than announcing an acute incident.
What hazmat cases teach about the genre as a whole
Two patterns:
Building-level granularity. Alerts in this chapter overwhelmingly name a specific building, often a floor or wing. Compare the active-shooter chapter, where alerts often must speak campus-wide because the threat is mobile. The hazmat genre has the easier composition problem: the threat doesn't move.
The closure is brief. Where the record preserves a hazmat all-clear, it tends to be short, a sentence or two that names the building and lifts the restriction, like the Rowan message above. The event ends as plainly as it began.
The hazmat cases are not the most cited in the literature on campus alerting, but they are the genre institutions most reliably get right. The case list below is a small library of plain institutional writing under genuine emergency conditions.