Chapter 36 — Key Takeaways

A scannable one-page card. For the full argument and the worked examples, see index.md.

The core claims

• The victim isn't always a person — but the logic is always the same. When the victim is an ecosystem, a structure, or the public, forensic science applies the identical reasoning it uses for a human victim: class vs. individual, exclusion over proof, validity, and bias.
• Wildlife forensics = the homicide lab's DNA, pointed at a non-human genome. Species ID is a class determination (often by DNA barcoding against a reference database); matching a tusk to a carcass reaches toward individual; geographic-origin DNA (elephants are the model) is an association with a place-type, the same logic as pollen (Ch. 13). Strong at its quantified core; limited mainly by database completeness.
• Environmental forensics attributes pollution by chemical "fingerprinting." GC-MS (Ch. 23) identifies what a pollutant is (strong, validated chemistry); attributing it to a source is a probabilistic comparison governed by the signature's distinctiveness and the completeness of the candidate set — like soil/glass (Ch. 24). The "fingerprint" metaphor overpromises uniqueness; a chemical pattern is a class characteristic.
• Forensic engineering / failure analysis = the scientific method applied to wreckage. Form hypotheses (design, manufacturing, overload/misuse, deterioration) and test each against affirmative physical evidence — fracture surfaces, deformation, materials testing, calculation. A cause needs affirmative evidence; concluding by elimination alone is the negative-corpus fallacy (Ch. 22) in a hard hat.
• Accident reconstruction = bloodstain analysis for vehicles. Physics turns skid marks, crush, rest positions, and EDR data into a speed range (a minimum, like the entomologist's PMI), not a speedometer reading. It constrains what happened; it does not narrate whose fault.
• Fire-and-explosion engineering (callback Ch. 22) can rule an accidental cause in or out. The key distinction is cause arcing (electrical activity that ignited the fire) vs. victim arcing (electrical damage produced by the fire). Mere melted wire proves nothing (every fire damages wiring); establishing — or excluding — an electrical cause requires affirmative analysis of which kind is present.

The method-validity verdict (NAS 2009 / PCAST 2016)

Where they sit: the quantified cores (wildlife DNA species ID, compound ID, calculable failure mechanics) are near DNA at the strong end; the attributions and reconstructions are probabilistic comparisons in the middle, strong only when distinctive and honestly bounded; eyeball conclusions and negative-corpus reasoning are the discredited folklore the book has condemned throughout.

What you can honestly say on the stand

• Wildlife (geographic origin): "The DNA of this tusk is consistent with, and supports, origin in [region]; it does not pinpoint a precise location, and its confidence depends on how well that region is represented in the reference map. It does not, by itself, identify who killed the animal."
• Environmental (source attribution): "The chemical pattern recovered from the spill is consistent with, and — given the distinctiveness of its weathering-resistant biomarker ratios relative to the candidates sampled — supports attribution to source A; source B is excluded. Unsampled sources cannot be ruled in or out."
• Failure analysis: "Based on the fracture surface, the reconstructed load path, and confirmatory calculation, the evidence supports failure by [fatigue / overload] initiating at [location]. I tested and eliminated the alternative causes, and this conclusion rests on affirmative evidence, not on the absence of another explanation."
• What you must NOT say: that a chemical "fingerprint" individualizes a source to the exclusion of all others; that a species ID from a partial, degraded sample is "definite" when a close relative could not be excluded; that skid marks prove recklessness or fault; or that "we found no accidental cause, therefore arson/defect."

Key terms (one line each)

• Wildlife forensics — forensic science applied to crimes involving wild animals/plants; centrally, species (and origin/individual) ID from biological evidence.
• Environmental forensics — determining the source, age, and responsible party of environmental contamination.
• Forensic engineering — engineering applied to investigating the causes of failures of structures, materials, components, and products for legal proceedings.
• Failure analysis — the systematic determination of how/why something failed, by testing hypotheses against the physical evidence.
• Accident reconstruction — physics and engineering applied to reconstruct how a collision/accident occurred, from the physical evidence it left.

The cold-case line

A forensic-engineering failure analysis ruled out the cabin's electrical system as an accidental cause of the fire — affirmatively (victim arcing, no competent ignition mechanism at origin), not by failure-to-find. It is an exclusion that complements the Ch. 22 incendiary finding from the opposite direction. Honest status: accidental-cause alternatives closed — no person named.

The themes this chapter advanced

• Exclusion over proof — wildlife DNA excludes the wrong species; environmental chemistry rules out a source; the engineer eliminates refuted hypotheses and rules out the electrical cause. The cleanest statement everywhere is not this one.
• The validity spectrum — every domain spans it: quantified cores near DNA, probabilistic attributions in the middle, eyeball/negative-corpus reasoning at the discredited bottom. The yardstick (tested, with a known error rate?) is identical for a tusk, a tank, and a truck.
• (Also advanced: cognitive bias — the retained engineer/reconstructionist told the wanted answer, §36.3–36.4; and the CSI effect — over-trusting a confident "chemical fingerprint" or dismissing a hedged engineering range, §36.6.)