When a wind farm slows its blades to protect bats, the story usually gets told as a before-and-after. Turbines sit idle a little longer into the evening, fewer bats meet a moving rotor, and the fatality count drops enough that everyone involved calls it solved.

That is roughly how curtailment gets described whenever it comes up, and for the most part the description is not wrong. It is just not the whole picture, and the case that put this fix on the map is a good place to see why.

The wind farm behind those Pennsylvania numbers is the Casselman Wind Project in Somerset County, a 23-turbine site that became one of the first places in North America where anyone tested this idea properly. Instead of letting all the turbines spin up the moment wind speed passed roughly 3.5 meters per second, the standard threshold, researchers led by Edward Arnett and Michael Schirmacher held twelve of the twenty-three back until wind speed reached either 5.0 or 6.5 meters per second, rotating each turbine nightly between full operation and one of the two curtailed settings so they could compare the same machines under different conditions rather than different machines under one condition.

Over the 75 days of nightly carcass searches that followed in the fall of 2008, field crews found 32 fresh bat carcasses at those twelve turbines. Read on its own, that number looks like proof the fix barely worked. It is not, and the reason is buried in how the experiment was actually run rather than in the headline count.

Of those 32 carcasses, 23 turned up on nights when the twelve turbines happened to be running at their normal, full-speed setting as part of the nightly rotation. Only nine were found on nights the blades had actually been held back, three under the 5.0 meter-per-second setting and six under 6.5.

Framed that way, 32 carcasses despite curtailment is really 23 carcasses on the nights curtailment was not switched on, and nine on the nights it was. That is a very different claim than the one the raw total suggests, and it is the kind of distinction that gets lost the moment a number gets lifted out of a methods section and into a headline.

What the reduction actually looked like

Once Arnett and Schirmacher compared nightly fatality rates instead of raw counts, the effect size was large. Full-speed turbines were killing bats at something like five times the rate of curtailed ones that season, and across the two years they ran the experiment, the researchers estimated that somewhere between roughly 72 and 82 percent of fatalities at those turbines happened on the nights the blades were spinning at normal speed rather than on the curtailed ones. Curtailment did not stop bats from dying at Casselman. It concentrated most of the deaths onto fewer nights and cut them sharply on the nights it was actually in effect, for a power cost the study put at around 2 percent of the output from those twelve turbines during the study window.

It is worth being precise about what kind of evidence this is. This was one study at one site over two fall seasons, but the design behind it is stronger than most single-site findings get credit for. Turbines were assigned to a treatment at random each night rather than singled out because they already looked dangerous, which is what lets you call the difference a real effect of curtailment instead of a coincidence tied to some other feature of the site.

It also is not an isolated result. Michael Whitby and four co-authors pulled together a decade of curtailment experiments at six wind facilities across North America and reported a strikingly consistent pattern. In a 2024 meta-analysis published in Ecological Solutions and Evidence, the team wrote: “We found that curtailment reduced total bat fatalities by 33% with every 1.0 ms−1 increase in curtailment wind speed.” At a 5.0 meter-per-second cut-in speed, the same relationship worked out to roughly 62 percent fewer fatalities on average, in the same range Casselman produced more than a decade earlier.

“Any small increase in this cut-in speed actually has a massive effect on bat survival,” says Emma Bennett, a wind and wildlife ecologist at Elmoby Ecology, who tested a far smaller adjustment at an Australian wind farm, moving the cut-in speed from 3 meters per second to just 4.5. Bat mortality there fell by around 54 percent, and the project gave up less than 0.1 percent of its revenue to do it. Different continent, different bat species, similar shape of result.

Why the small number still matters

None of that makes curtailment a fix in the sense that a quick headline about raising cut-in speed to protect bats implies. It reduces risk substantially and repeatedly across sites, seasons, and species, which is a genuinely strong track record for a wildlife mitigation measure. It does not zero out the risk, and a wind farm that raises its cut-in speed and then stops asking questions is not off the hook for the bats still dying on the nights curtailment happens to be running, let alone the ones on the nights it is not.

I am not a wildlife biologist or an energy engineer, and I do not have any standing to referee the finer points of turbine aerodynamics or bat flight behavior. What stopped me on this story was something more familiar to me: the instinct to treat a fix as either a total solution or a failure, with nothing honest in between. I catch myself doing that in my own life more than I would like to admit. I will look at some habit or system that is not working well and either overhaul the entire thing or write it off as not worth touching, instead of asking whether a modest, cheap adjustment would already get me most of the benefit. Not every problem needs the maximal intervention, and I think people, myself included, often assume it does because a partial fix feels less satisfying to report on than a clean before-and-after.

Curtailment is a good example of a fix that earns real credit for what it does, which is cut deaths substantially at a cost of a percent or two of energy output, without deserving credit for what it does not do, which is eliminate the risk entirely. Both of those things are true about the same number, and treating them as a contradiction is what makes a study like the one at Casselman look, at a glance, like it failed.

What a number like 32 is actually telling you

The next time a headline pairs a mitigation measure with a body count, the number worth asking about is not the total. It is what fraction of the exposure that measure actually covered. A wind farm that curtails on some nights and not others, that curtails some turbines and not the rest, or that curtails only after a certain wind speed instead of shutting down entirely, is choosing a point on a curve between full harm and full protection, not claiming to have found the top of it. Casselman’s twelve turbines were never going to produce zero carcasses over 75 days of searching, curtailed or not. The question that actually matters is whether the deaths that did happen were concentrated on the nights the mitigation was not running, and at Casselman, and in the decade of studies that followed it, they mostly were.