I can't say that Susan was particularly excited by my latest leanings toward taking firefighting class. I believe her immediate response was "I don't want to spend years taking care of a charred and paralyzed person."

I admitted that the most mutually respectful approach would be to reach some sort of an accord, a compromise. I asked first if there was anything that would help allay her fears.

She answered, without a moment's pause: "Local mortality data from the past few years, demonstrating that you'd be at no greater risk than you're already at, as an EMT."

I had little doubt that such data, should I locate it, would substantially weaken my case. But I hope to never shrink from the truth, so I went looking for some statistics on firefighter deaths. Of course, mortality is less pertinent than morbidity, since Susan is specifically concerned about me becoming horrifically disabled. She wouldn't be pleased if I died, either, but she would stand to collect a decent amount of life insurance, which might help soften the blow. In any case, I could only find federal records, covering firefighters' on-the-job mortality through 1994. They were fascinating. For example, the number one cause of on-the-job death among firefighters? Heart attack. Burns, which I expected to be the flat-out prime killer, lagged behind even gunshot wounds. I did not find that last fact to be particularly reassuring.

I quickly shared the news about heart attacks with Susan, as it suggested I might be relatively safe; I don't present as someone likely to be felled by heart attack in the immediate future. Then I started crunching some numbers. In 1994, 95 firefighters died in the line of duty. Considering that there are around 1,080,000 career and volunteer firefighters in the nation ('96 data), that doesn't make firefighting seem like an utterly ridiculous risk (.0088%). But then I flipped the equation over: one in 11,300 firefighters dead every year. A one in 11,300 chance that it will be you (me). That seems like a lot. It's probably a greater risk of dying on the job than that of, say, freelance website designers. Or volunteer EMTs.

Of course, that's a blatant misuse of math: it presumes that risk is spread evenly throughout the entire population of firefighters. Instead, I would expect to see much higher levels of risk in busy urban areas than in rural locations (or, higher risk in rural areas because they run fewer calls and are therefore less experienced in a real fire). A more illuminating analysis would have been number of deaths by number of calls run. If number of calls accounted for significant variance in the mortality rate, then I could roughly assess the risk factor based on how many calls I would run. Of course, some other, unexpected factors might account for substantial portions of the variance in mortality rates. Wouldn't it be fun to sink your hands into all that raw data and just dig around in it with a nice, powerful piece of statistics software? Imagine the little gems of hidden truth that lie beneath that untouched ground.

I took temporary solace in the capacity for logic, math, and powerful statistics software to reduce the ridiculous world to fewer, more manageable components. But in spite of the numbers, I still feel the strange pull to a shimmering lure that I first saw in the room with the blinding smoke. And Susan, though vowing to look at the statistics, is driven by the same kind of deep emotion that I am, the kind we cannot fully explain but that shows us something true. We both know the numbers don't matter too much. We're swiftly moving past the realm of statistics. The conversation we're having makes math look like pig latin.