The part of a fuel spill you can see is almost never the part that kills anything.
A sheen on the surface looks like the disaster. It is mostly the heavy end, sitting on top, visible and easy to skim.
The damage is done by what dissolves.
Gasoline carries a light aromatic fraction that goes into water readily, spreads through the whole depth of a stream, and passes across a gill without any effort at all.
By the time responders arrive it is already in solution and already moving.
Which is why a small volume in a small stream can be a total loss.
The fuel that hides under the streambed
There is a second problem, and it is the one that decides how long a cleanup runs.
A creek bed is not a pipe. It is loose gravel with water moving slowly through the spaces inside it, sometimes a foot or more below the visible channel.
Fuel works its way down into those spaces and stops moving. Skim the surface and the creek looks clean while the gravel underneath keeps releasing it back.
So crews attack the bed rather than the water. They flush sections deliberately to drive trapped product out, and they push air through the gravel so the light ends lift off as vapor instead of staying dissolved.
That second technique is why the response involved compressors rather than only absorbent pads.
Each pass leaves a little less behind, and the work ends when the readings stop climbing back up.
A creek that feeds a rebuilt river
The stream in question is a tributary on the north edge of the Olympic Peninsula, joining a river that has become the largest dam removal story in the country.
Two dams came out of that river, and salmon have been working their way back into water they could not reach for a century.
That is the context that makes a few thousand gallons a serious event.
The same water also supplies a city system, and residents on it were told not to drink from the tap while the first samples were run.
That advisory was lifted the same evening once testing came back clean.
The creek itself is small, shallow and cold, which is exactly what juvenile fish want.
What the count came to
The truck left US Highway 101 on July 18 of 2025 and went down toward the water.
The recorded release was 2,798 gallons of gasoline and 172 gallons of diesel, a little under 3,000 gallons in total.
Wildlife teams then walked the affected reach and counted dead fish by hand, arriving at 2,134 animals.
Those were not all salmon. The tally included juvenile salmon along with lamprey and sculpin, which are the species that live in the gravel and cannot leave it.
Containment boom went in at four separate points, absorbent material was swapped daily, and air and water monitoring ran continuously through the response.
A tribal government, a county sheriff, a fire district, a state agency, a federal one and the trucking company all sat in one command structure, which is standard practice and rarely this crowded.
The number is a floor and not a total
A hand count of dead fish in a creek is the best available figure and it is not the real one.
Fish that die in moving water get carried off, eaten, or lodged under banks where nobody finds them, so a walked survey always reports fewer than actually died.
It also says nothing about eggs already in the gravel, or about the insects that the next year of fish were going to eat.
Damage assessment on a spill like this runs for years afterward and is a separate process from the cleanup, in the way that a hatchery truck rollover or runoff from a road leaves a bill that arrives years afterward.
The state agency published its own account of the sparging and flushing work, describing the effort to save the creek as months rather than days.
Why the fish were let in anyway
Here is the decision that sounds wrong and was not.
Adult salmon were due back to spawn while the creek was still being worked on, and the command could have blocked them at the mouth.
Blocking them means an entire year class never reproduces, in a run that is only just recovering, and there is no way to get that year back.
Letting them up means some of them spawn in water that is cleaner than it was and dirtier than it should be, and most of the eggs probably make it.
The federal damage program still lists the case as open, with restoration yet to be settled.
The returning salmon were not protected from the spill.
They were protected from a lost season.
Read the whole thing?
Get the week's signal, not the noise
Our sharpest reporting on energy, climate and nature — free, once a week.
