But a few paragraphs later, we learn that America lost 33,000 jobs from August to September, far worse than the increase of around 80,000 the experts forecast, and a severe reversal from the average monthly gains of 172,000 over the past year.

So I may be speaking for broad audience when I ask, how can it be that the jobless rate improved sharply, while jobs shrank by the biggest number since July 2001?

The answer is that the two numbers come from separate surveys that measure different things. Both were conducted during the same week in mid-September. The Household survey, which questions America’s families, uses methodology that measures the labor market’s longer-term, overarching trends, while the Business survey, which polls employers, provides a snapshot that highlights the impact of wrenching or serendipitous events that can radically shift payroll numbers from month to month.

In September, the two measures pointed in radically different directions. But a dive into the details, guided by Mark Hamrick, senior economic analyst at Bankrate.com, the shopping site for financial services, clears the confusion. In reality, the employment picture for September remains excellent. “On balance, it was a good report,” says Hamrick. “What counts most is that a lot more people supporting families are working this month than last month.”

First, let’s look at the impressive Household numbers. Hurricane Irma struck the precise week BLS was conducting the Household polling, and Harvey had hit three weeks before, so its damage was still fresh. But big disasters, no matter how severe, barely budge the Household findings. That’s because it counts all people who still have jobs, but aren’t paid during the week of the survey, as still employed. A ferocious storm that closes restaurants and caterers for a few weeks, but doesn’t actually eliminate the jobs of the people who work there, doesn’t count.

In fact, the overall picture of strongly growing payrolls not only remained intact, it actually improved. From August to September, the ranks of the unemployed shrank by 330,000, driving the jobless rate down an impressive 0.2% from what was already one of the lowest levels in modern history. “We tend to breeze past those numbers, but 330,000 is the population of a mid-sized city, and those wage-earners are often supporting families,” says Hamrick.

The Business survey counts everyone who was getting paid last month, but didn’t get a paycheck in this month’s survey, as a lost job. By that measure, the economy indeed “lost” 33,000 jobs, driven by a steep fall of 105,000 at food and drinking establishments, businesses that had been adding 24,000 positions a month.

Hamrick reckons that the hurricanes have caused a temporary, downward blip in job-creation numbers, but that the sunny outlook remains in place. “We’re seeing worker shortages in construction, and better growth in mining and logging, two industries that the Trump administration supports,” he says. “And the strength in service more than offsets the efficiencies in automation and tech that restrict job growth in manufacturing.”

Truth be told, this reporter (whose specialties run to other esoteric topics like tax reform and corporate governance) never knew the two surveys could seem to contradict one another. But what they really do is tell different stories, one a narrative, the other a moment in time. And the former shows that hurricanes can’t dent the resilience of America’s resurgent jobs machine.