The Guard That Came Off

In the fall of 1988 I was nineteen and working nights on the maintenance crew at a wire mill in Kewanee. Around two in the morning an operator I’ll call Ray was threading a new coil through a drawing block when his glove caught. I was the closest person to the stop button. I hit it, and it was too late for two of his fingers.

The injury is only part of what I remember. The rest is what people said while we stood around the block waiting for the ambulance. The block had a guard. The guard had come off months earlier because it slowed down thread-up. Everybody on that shift knew it was off, including the supervisors, and including me. Nobody decided that Ray should get hurt. We had all decided, one shift at a time, that the shortcut was fine.

I’ve spent a good part of my working life since then trying to understand how a crew of careful, decent people gets to that point together. I haven’t finished. I have learned a few things.

Small tolerances

The serious injuries I’ve seen all had a history. A guard that was a nuisance. A forklift lane that cut across a walkway. A floor that got slick near the galvanizing line when it rained. Each one was minor, and each had been there long enough that nobody saw it anymore.

So most of my job on safety is to keep noticing those things, and to make noticing them normal for everybody else. Speeches don’t do much.

Huntington

When I took over a fastener plant in Huntington, Indiana, in 1999, its recordable injury rate was close to nine. Roughly nine people in every hundred got hurt badly enough each year to need more than first aid. The plant had safety posters, a safety committee and a monthly safety meeting. It also had a culture where reporting a near miss felt like turning in a coworker.

We changed three things. Safety became the first item at every meeting, starting with the shift huddles at the machines, and every supervisor started each shift by walking their area looking for hazards. Near-miss reports could be anonymous, and we fixed what was reported within a week or explained in writing why we couldn’t yet. And anyone in the building could stop a line if something looked wrong.

The third one took the longest. People didn’t believe it until an operator stopped a line on a Friday afternoon with a big order due, and I thanked her by name at the next all-hands. Four years later the injury rate was 2.4.

Counting the right things

A recordable rate tells you about injuries after they happen, and it is easy to game. Pay bonuses for low injury counts and the counts will go down. The injuries may not.

At Cordell we stopped tying any bonus to injury numbers in 2016. We track these instead:

• Near misses and hazards reported each month, per hundred people. We want this to go up.
• The share of reported hazards fixed within seven days.
• Supervisor safety walks completed, and what they turned up.
• Lockout and guarding re-qualification, tracked by person.

When we started, people reported about four hazards a month per hundred employees. Now it’s over twenty. Over the same stretch our recordable rate went from 5.6 in 2014 to 1.2 in 2025, and in July the Jackson plant passed a million hours without a lost-time injury. I don’t think the two trends are unrelated.

2023

None of this means we’re finished. In 2023 an operator in Danville slipped near the galvanizing line on a wet morning and broke her wrist. It was one of those small, known hazards. The floor drain in that corner had been on the maintenance list longer than it should have been. We fixed it within the week, about eight months late.

She was back at work in two months. She also offered to walk our board through what happened, standing at the spot where she fell. The board asked better questions that afternoon than they had in any safety review I’d given them.

I still think about Ray. He went back to work at the mill and retired from there years later. The guard was back on that block by the start of the next shift.