In July 1988, a thirty-one-year-old journeyman electrician was installing a flood light on a new house in Virginia. He had been in the trade two years and with that contractor four months. The company was a multistate electrical contractor with a written safety program, and its foremen ran weekly tailgate safety meetings. None of that is offered as irony. It is offered because it is what the investigators found, and because it rules out most of the explanations people reach for first.
He was about eighteen feet up an aluminum extension ladder, stripping insulation from the house wiring with insulated strippers. His right thumb and index finger contacted the uninsulated part of the tool. The circuit had not been de-energized at the panel. The aluminum ladder never touched the conductor; it provided the path to ground. He died of electrical injury. The investigation is NIOSH FACE 88-37 and it runs to three pages.
What the investigators wrote down
Two findings in that report are worth sitting with.
The first: NIOSH recorded that it is not known whether the victim or the foreman realized the circuit was energized. Not that somebody took a risk and lost. Not that somebody was careless. It is not known — which is to say that on that job, on that day, the state of that circuit was a matter nobody could establish with certainty even afterward.
The second follows directly from it. The panel box containing the circuit breakers was not labeled.
The directory is a safety device
Article 408.4(A) requires every circuit in a panelboard to be legibly identified as to its purpose, marked at the panel, and specific enough that somebody can tell which circuit is which. Article 110.22 requires disconnecting means to be legibly marked to indicate their purpose. Between them they describe a piece of paperwork that gets treated as a courtesy to the next guy, filled in with a pencil if there is one in the truck, and skipped entirely on the rough because it can be done at trim.
Look at it through this case and it is not paperwork. It is the mechanism by which the circuit you are about to work on becomes lockable. Lockout starts with identification. If nobody can say which breaker feeds the flood light circuit, then the sequence that keeps you alive has failed at step one, and everything downstream of it — the lock, the tag, the verification — is being done on a guess.
This is also why the question turns up constantly from people who are not electricians: how do I safely work out which breaker this is. The honest answer is that in a properly identified panel it takes a few seconds, and in an unlabeled panel it takes a partner, a load, and patience — and that the reason the code asks for the directory is precisely so that the hard version is not the normal version.
Identification gets you to the right breaker. It does not tell you the conductor is dead. That requires a test, and the test has a procedure that is worth being pedantic about: prove your tester on a known live source, test the conductor, then prove the tester again on the known source. If you skip the third step you have no way of knowing whether your instrument died between the first reading and the second.
This is where the long-running argument about non-contact voltage testers actually lands. They are convenient and they are useful for a first pass. They are also capable of a false negative — through certain jacketing, in certain geometries, with a low battery, in a shielded raceway — and a false negative from a tester is worse than no tester at all, because it converts a suspicion into a confidence. For anything you are about to put your hands on, contact test it.
The ladder, and the thing that was not there
The report notes that the aluminum ladder did not contact the energized conductor. It simply gave the current somewhere to go. That distinction matters because it is the one people get wrong when they retell this kind of story: the hazard was not that the ladder touched something, it was that the ladder made the man a viable path to earth. A fiberglass ladder would not have removed the fault. It would have removed the circuit.
The other absence is harder to see. On a construction site, ground-fault circuit interrupter protection for personnel on temporary power is an OSHA requirement, set out in Subpart K of the construction standards, and it exists for exactly this scenario — somebody in contact with an energized part and a path to ground, on a job where the wiring is half finished and nothing can be assumed. A GFCI does not care whether the panel was labeled or whether anybody believed the circuit was dead. It cares about current going somewhere it should not, and it acts in milliseconds.
It is also worth noting what the contractor did have. A written safety program. Weekly tailgate meetings. Those are not nothing, and the investigators recorded them. But a program is a set of intentions, and what failed here was a physical control: a labeled panel, a lock, a verified test. Paper safety and physical safety are different things, and the gap between them is where most of these reports live.
Why this case, and not a statistic
The numbers exist and they are not small. BLS counted 142 fatal injuries from exposure to electricity in 2024, up from 130 in 2023, and the direction of travel is the wrong one. Figures like that tell you the trade is dangerous. They do not tell you anything you can act on tomorrow morning.
A single case does. FACE 88-37 describes a competent man, two years qualified, working for a company that did the training, on an ordinary task, who was defeated by an unlabeled panel and an assumption. There is no dramatic failure in it anywhere. That is what makes it useful. NIOSH keeps a searchable database of these investigations, each one written up with the sequence of events and the recommendations, and reading a few of them changes how a rule reads. The rule stops being an inconvenience somebody invented and becomes the summary of something that already happened to a person who was probably better at this than you assume.
Label the panel. Prove the tester twice. Neither costs anything.
Our illustrated guide to understanding the NEC, volume one tries throughout to explain where requirements like these came from, because a rule you understand the origin of is a rule you stop skipping when the schedule tightens.