Phase 3 Soil Results: What They Show, and What They Do Not Explain
The Phase 3 residential soil reports have been shared. They confirm PFOA in garden soil across the tested area. They also raise plain questions the reports do not fully answer: the contamination is directional rather than evenly spread, the highest readings are to the south-east, and the model used to choose where to test has not been published.
- WSP Phase 3 Residential Soils Investigation (March 2026), prepared for the Environment Agency, ref 70100310/12294. Shared via Leigh Day.
- WSP Phase 2 Factual Report (2025) — peak PFOA 144 µg/kg at GR01-SP232.
- Dutch (RIVM) soil screening value for PFOA: 3.8 µg/kg.
- EA AQMAU-C2478-RP02 (2023), air dispersion and deposition modelling study — cited in the Phase 3 report; not published.
The Phase 3 residential soil reports have now been shared. This post sets out what they show in plain terms, and the questions they leave open. It is limited to what the documents themselves say.
What was tested
Phase 3 looked for PFOA in garden soil at selected residential properties near the Hillhouse site. PFOA is one of the PFAS chemicals made at the site for many years. The investigation focused on PFOA reaching gardens through the air and settling on the ground.
For context, the Dutch screening value for PFOA in soil is 3.8 µg/kg. That is the level above which the Dutch authorities consider soil worth a closer look.
What the results show
PFOA was found almost everywhere it was tested. In Phase 2, PFOA was detected in 98% of samples (204 of 208), with a peak of 144 µg/kg. Phase 3 extended testing into residential gardens and detected PFOA in 99% of its samples, with readings up to 181 µg/kg and an average across the combined dataset of about 5.9 µg/kg. For a contaminant in garden soil, that is close to universal across the area near the site.
The pattern is worth pausing on. The single highest reading, 181 µg/kg, was recorded to the south-east at around 700m from the site, higher than the 144 µg/kg peak at 500m. The report says the 144 reading sat at the outer edge of the earlier 500m survey and indicated a potential for deposition reaching beyond it. It also finds no clear correlation between concentration and distance, which it puts down to the spread of sampling points and the airborne nature of the deposition. So the contamination does not simply fade with distance, and how far it reaches beyond the tested area is not established in the documents.
The questions the reports do not answer
Were all the affected directions tested? The report’s own measured data shows the highest PFOA concentrations in two directions from the site: the south-east (around 120 degrees) and the west-north-west (around 280 degrees). Phase 3 concentrated its testing to the south-east. Whether properties in the west-north-west direction were tested as thoroughly is not clear from the documents.
Why can the public not read the model behind the testing? The testing locations were planned using the EA’s dispersion model, AQMAU-C2478-RP02 (2023). That model is referred to in the report but has not been published. The wind data behind it, the assumed stack height, and the emission figures are not set out in the material available to residents. Without them, the basis for where testing was focused cannot be checked.
What about the other PFAS? Phase 3 looked at PFOA. The site is associated with more than one PFAS chemical. Where other PFAS appear in the data, the reports do not investigate where they came from.
Can a resident ask for their own garden to be tested? The documents do not describe a route for a resident to request testing of their own property, or set out the criteria used to choose which properties were tested.
What is not confirmed in available documents
These are questions, not conclusions. The reports do not state that contamination stops at the tested boundary. They do not explain the choice of testing area against the measured results. And the model used to plan the testing is not available to read. Each of these is a reasonable thing to ask for, and we have asked Leigh Day to request the underlying data and the dispersion model in full.