Chapter 21: Forensic Chemistry: Drug Identification, Explosives Analysis, and Fire Investigation

"The screening test gives you a reason to keep working. It was never meant to be the verdict — and most of the trouble in this field begins the moment someone treats it as one." — constructed teaching epigraph, in the voice of this book

Overview

A patrol officer finds a plastic bag of white powder in a glovebox. A bomb technician sifts a crater for the few milligrams of unconsumed material that survived a detonation. A fire investigator scoops charred flooring into a clean can because the room smelled, faintly, of gasoline. Three very different scenes, three very different stakes — and underneath all of them, one question the chemistry bench exists to answer: what is this substance? Not who handled it, not what it means for the case, not whether a crime occurred. Just: what is it, and how sure can we honestly be?

This is the chapter where forensic science is, for once, on its firmest ground — and where the danger is not that the science is weak but that the fast version of it gets mistaken for the certain version. Analytical chemistry is genuinely rigorous; a modern crime lab can identify a controlled substance, an explosive, or a fire accelerant with a confidence that DNA's statistics would recognize as kin. But every one of those identifications has two stages, and they are not equal. There is the presumptive stage — a color that blooms in a test tube, a crystal that forms under a microscope, a handheld device that beeps at the roadside — which is fast, cheap, and suggestive. And there is the confirmatory stage — the instruments of Chapter 23 — which is slow, expensive, and definitive. The whole discipline of forensic chemistry, and most of its failures, live in the gap between those two words.

We will build the chapter around that gap. First, what the chemistry section does and the substances it handles. Then controlled-substance analysis — the color and microcrystalline tests that screen a suspected drug — and the presumptive/confirmatory pairing that should govern every result. Then explosives and their residues, where the sample is often nearly gone before the analyst sees it. Then fire-debris chemistry, where we will set up the arson reckoning of Chapter 22 and the instruments of Chapter 23 — and where the cold case adds its next piece. And finally, the chapter's moral center: the roadside field test, a presumptive color reaction that, used as if it were an identification, has put demonstrably innocent people in jail. Two themes run through all of it: not all methods are equally valid — presumptive suggests, instrumental confirms, and the CSI effect and the courtroom both reward overstatement. By the end you will know what a chemist may honestly say after a color test, and the exact words that would be a lie.

In this chapter, you will learn to:

• Define forensic chemistry and controlled-substance analysis, and say what the chemistry section is actually asked to do.
• Explain how a presumptive color test and a microcrystalline test work, and what each can and cannot establish.
• State the presumptive/confirmatory pairing as the bench's organizing logic, and why a screen is a lead, not an identification.
• Describe how explosives residue is detected and analyzed, and the contamination problems that bound it.
• Explain fire-debris chemistry and why an indicated accelerant is not a confirmed one (previewing Chapters 22–23).
• Diagnose the roadside field-test problem and place these methods on the NAS 2009 / PCAST 2016 validity spectrum.

Learning Paths

🔎 Investigator/CSI: Your decisions at the scene determine whether the chemistry is even possible. Sections 21.4 and 21.5 are yours: how a fire-debris sample or explosives swab is collected and packaged (a nylon bag, a clean unused paint can, a control sample) without destroying or contaminating the trace. And §21.6 is a warning you carry on your belt: the field test in your hand is a screen, and an arrest built on it alone is a risk you should understand. 🧪 Lab analyst: Weight 21.2, 21.3, and 21.5. The color and microcrystalline chemistry, the discipline of the two-stage funnel, and the handoff to the instruments (Chapter 23) are your bench. Know exactly what each screening result licenses you to write — and what it does not. ⚖️ Law/courtroom: Sections 21.3 and 21.6 are where cross-examination lives. The gap between "presumptively positive" and "identified," the legal weight a confirmatory result carries that a screen does not (recall Melendez-Diaz, Chapter 5), and the documented wrongful arrests from unconfirmed field tests are the heart of your work here. 👥 General reader/juror: §21.1 and §21.6 are the antidote to the television version, where a powder is "tested" once and instantly named. Watch how a real identification is built in two stages, and how much harm follows when the first stage is treated as the last.

21.1 What the chemistry section does

Begin with the question the bench is asked, because the chemistry section's strength comes precisely from how narrow and answerable that question is. Forensic chemistry is the application of analytical chemistry to legal questions — chiefly the identification and characterization of substances whose identity matters to an investigation: controlled substances, explosives and their residues, fire accelerants, and unknown materials of every kind. Notice what is not in that definition. The chemist identifies the powder; the chemist does not decide whether possessing it was a crime, who owned the glovebox, or what should happen to the defendant. The substance is the witness, and the chemist's job is to make it speak its name accurately, and to refuse to make it say anything else.

That restraint is the source of the field's rigor. Recall the validity spectrum from Chapter 1: instrumental forensic chemistry sits near the strong end, alongside DNA and unlike bite marks, for a specific reason. Its core claim — "this substance is X" — is grounded in two centuries of analytical chemistry, it is testable, it has measurable error rates, and its instruments (Chapter 23) produce a molecular result that can be checked, re-run, and defended. When a chemist says, after a full instrumental analysis, "this is cocaine," that statement rests on the same kind of validated foundation as a single-source DNA match. The field earns its place on the spectrum honestly.

But — and this is the whole chapter — that strong identification is the end of a process, not the beginning. The chemistry section works the same two-stage funnel you met in serology (Chapter 10): a fast, cheap, sensitive screen that triages and suggests, followed by a slow, specific confirmation that identifies. The screen alone is a different and weaker kind of statement than the confirmation, and almost everything that goes wrong in forensic chemistry goes wrong because the two are confused — a roadside color change reported as if it were the instrument's verdict.

🔬 At the Bench A working crime lab's chemistry section, or "drug chem" unit, is one of its busiest. The casework is dominated by suspected controlled substances — the single largest category of evidence in many American labs by sheer item count — which is exactly why screening exists: an analyst cannot run a half-hour instrumental method on every one of thousands of submitted baggies. So the workflow is triage first. A typical sequence: visually and physically examine the item (color, form, packaging, any logos on pills); weigh it (the quantity often determines the charge, so the balance is a forensic instrument too, calibrated and logged); run one or more presumptive tests to decide what the substance might be and which confirmatory method to choose; then run the confirmatory instrumental analysis that actually identifies it. The screen's purpose is to direct the confirmation, not to replace it.

The substances the section handles fall into a few broad families, and the rest of this chapter walks the three most consequential. Controlled substances — illegal drugs and regulated pharmaceuticals — are the daily bread (§21.2–21.3). Explosives and their residues are a smaller, specialized, high-stakes category, where the sample is often almost entirely consumed by the event being investigated (§21.4). Fire debris — charred material that may carry the chemical ghost of an accelerant — is the bridge to the arson chapter (§21.5). All three are united by the funnel, and all three teach the same lesson from different angles: the screen suggests, the instrument confirms, and the honest chemist never lets the first wear the clothes of the second.

21.2 Controlled-substance analysis: color and microcrystalline tests

Now to the daily work. Controlled-substance analysis is the forensic identification of drugs and regulated pharmaceuticals — determining what a seized substance is, and often how much of it is present. It is the highest-volume work in forensic chemistry, and it begins, almost always, with one or both of two classic presumptive techniques: the color test and the microcrystalline test. Understand these two and you understand the screening stage of the entire field.

Color tests: chemistry you can see

A presumptive color test is a chemical screening test in which a reagent produces a characteristic color — or a sequence of colors — when it contacts a particular class of substance, indicating that the substance is probably present and, just as usefully, indicating where it is probably not. The chemistry is straightforward: the reagent reacts with some functional group common to a drug class, and the reaction product is colored. The analyst places a tiny amount of the unknown in a spot plate or test tube, adds drops of the reagent, and reads the color against a known expectation.

Several of these reagents are old, named, and standard. You do not need to memorize them, but a few anchor the idea:

• The Marquis reagent turns a vivid purple-to-black in the presence of many opioids (heroin, morphine) and a distinctive orange-to-brown with amphetamine-type compounds. It is the most widely used general screen.
• The cobalt thiocyanate family (the Scott test for cocaine is the classic) turns blue in the presence of cocaine.
• The Duquenois-Levine test produces a purple color used to screen for cannabis (marijuana).
• The Mecke and Mandelin reagents give their own color signatures and are used alongside Marquis to narrow a class.

   A PRESUMPTIVE COLOR TEST — WHAT IT DOES AND DOESN'T SAY
   (illustrative reagent → color associations; not a reference chart)

   UNKNOWN  ──drop reagent──►  COLOR DEVELOPS  ──►  "consistent with a CLASS of substance"
   powder                       (e.g., Marquis → purple-black)
                                      │
                                      │  but the SAME color can come from:
                                      ▼
                          ┌───────────────────────────────────────────┐
                          │ • a different drug in the same class       │
                          │ • an unrelated compound that reacts alike  │
                          │ • a legal substance with the same group    │
                          └───────────────────────────────────────────┘

   POSITIVE = "screen consistent with class X — CONFIRM instrumentally (Ch. 23)"
   NEGATIVE = "class X not indicated here" (often the more reliable result)

Read that diagram the way you read the serology funnel in Chapter 10, because the logic is identical and the negative is again quietly the stronger result. A color test that fails to develop the expected color is decent evidence the screened-for class is absent — a reliable-ish exclusion, the kind of "no" the book keeps prizing (Theme 1). A color test that does develop the color tells you only "consistent with this class — keep going." It cannot tell the members of a class apart (Marquis purple does not distinguish heroin from morphine from codeine), and worse, substances outside the class can produce the same or a confusingly similar color. The color is a hypothesis, not an identification.

⚠️ Junk-Science Alert The single most dangerous sentence in this chapter is "the field test came back positive for cocaine, so it's cocaine." A color test does not identify a specific compound; it indicates consistency with a class, under conditions where many other substances — some utterly legal — can trigger the same color. There are documented instances of common, innocent materials producing "positive" color reactions on roadside kits. A color is a reason to do confirmatory chemistry. Treated as a conclusion, it is precisely the kind of unvalidated, overstated claim Chapter 1 taught you to flinch at — and, as §21.6 shows, it has put innocent people in cells.

Microcrystalline tests: shapes under glass

The second classic screen is subtler and, in skilled hands, more discriminating. A microcrystalline test is a presumptive technique in which a reagent is added to a trace of the suspected substance on a microscope slide, causing the target compound to form crystals of a characteristic shape, color, and habit that the analyst identifies under the microscope. Different drugs, reacted with the right reagent, throw down crystals of recognizably different geometry — needles, rosettes, plates, dendritic branching forms — and an experienced analyst can use those shapes to narrow an identification more tightly than a color test allows.

Microcrystalline tests are more specific than color tests, require very little sample, and for some substances (cocaine is the textbook case) are quite reliable in expert hands. But they remain presumptive, and for honest reasons: the result depends on the analyst's training and judgment in reading crystal morphology (a subjective element, with the bias risks that subjectivity always carries), crystal habit can be affected by impurities and cutting agents, and the method does not produce the molecular-level certainty an instrument does. They screen well; they do not, on their own, confirm.

🔬 At the Bench Why keep using presumptive tests at all, if they cannot identify? Because triage is not optional. A lab receiving thousands of drug items cannot put each one through a thirty-minute instrumental run sight-unseen; the screen tells the analyst which confirmatory method to choose and flags the rare item that is not what it appears. Presumptive tests are also cheap, fast, portable, and need almost no sample. The error is never using them — it is stopping at them. The bench's discipline is to treat every presumptive positive as an unanswered question routed to the instrument, and to write the report in the language of "consistent with," never "is."

Why the screen fails: a moving target

There is a further, modern reason the screening stage cannot be trusted to identify, and it has grown more acute in recent years. Color and crystal tests were developed against a relatively small, stable menu of drugs — the classic opioids, cocaine, the amphetamines, cannabis. The illicit drug supply is no longer stable. A flood of novel psychoactive substances (NPS) — synthetic cannabinoids, designer stimulants, and above all the proliferating family of fentanyl analogs — now appears and mutates faster than reference data can keep up. These present two distinct problems for a presumptive screen. First, a new analog may give no color reaction, or an unexpected one, and so be missed or misclassified by a test designed for an older compound. Second, and more dangerous, drugs are increasingly encountered as mixtures and adulterated products — a powder cut with multiple active and inactive components — and a color test reads the mixture as a blur, unable to resolve that, say, a "heroin" sample also contains a fentanyl analog at a potency the screen cannot flag. The class-level coarseness that always limited color tests becomes a genuine public-safety gap when the unidentified component is lethal in microgram quantities. The lesson reinforces the chapter's spine from a new direction: the screen tells you less, not more, exactly when the stakes are highest, and only an instrument that resolves and identifies each component (Chapter 23) can keep pace with a supply designed, in part, to defeat older tests.

21.3 The presumptive/confirmatory pairing

We have now met the two halves twice — in serology (Chapter 10) and in drug chemistry — so let us state the principle in full, because it is the backbone of this chapter and the thread that Chapters 22 and 23 hang on. Every confident forensic-chemistry identification is the product of a pairing: a presumptive test that screens and a confirmatory test that identifies. The two are not interchangeable, not redundant, and not equal in court. They answer different questions, at different strengths, and the entire integrity of a drug, explosives, or arson conclusion depends on keeping their roles straight.

Here is the distinction, drawn sharply:

The decisive column is the last one. A presumptive result is investigative information — enough to seize an item, to direct the lab's choice of method, sometimes to support a probable-cause arrest — but it is not, by the standards of forensic science, an identification, and most laboratory and professional standards forbid reporting a substance as identified on a screen alone. The confirmatory result is the one that belongs in a final report and on the witness stand. When you hear a chemist testify that a substance "is" a controlled drug, the honest version of that sentence is backed by an instrument; the dishonest version is a color test wearing a lab coat.

Why does confirmation require an instrument, specifically? Because of a principle the field calls orthogonality: a sound identification combines techniques that are based on different physical principles, so that the ways they could each be fooled do not overlap. A color test and a microcrystalline test, run together, are better than either alone — but both are, broadly, chemical-reaction screens, and a substance that fools one chemistry may fool a related one. An instrumental method like mass spectrometry (Chapter 23) interrogates the molecule itself — its mass and the way it fragments — by a completely different physical route. When a screen and an orthogonal instrumental method agree, the chance that both were fooled in the same way drops toward zero. That convergence of independent methods is what turns "probably" into "is," and it is the same logic of converging, independent evidence that the whole book builds toward at the capstone (Chapter 39).

⚖️ In the Courtroom The law has, in places, caught up to this science. In Melendez-Diaz v. Massachusetts (Chapter 5), the U.S. Supreme Court held that a forensic laboratory's certificate identifying a substance as a controlled drug is testimonial — the analyst who performed the work can be required to appear and be cross-examined, rather than letting a bare certificate stand. The practical effect for this chapter: an analyst on the stand must be able to say which tests were run and defend why they support the identification. "We ran a presumptive color test" is not a sustainable answer to "how do you know this is heroin?" The competent cross-examiner's question is always the orthogonality question — what confirmed it, and by what independent principle? — and a case resting on a screen alone does not survive it.

🔍 Check Your Understanding 1. A report states a powder "is methamphetamine," but the case file shows only a Marquis color test was performed. Name the specific overstatement, and write the sentence the report should contain instead. 2. Why are a color test plus a microcrystalline test together less convincing than a color test plus a mass-spectrometry result, even though the second pairing is only two methods? (Hint: orthogonality.)

21.4 Explosives and their residues

Explosives analysis is forensic chemistry under the hardest possible sampling conditions, because the very event under investigation tends to destroy the evidence of itself. When a device functions, most of the explosive is consumed in the reaction; what the analyst gets is whatever survived — trace quantities of unreacted material, characteristic decomposition products, and fragments of the device — scattered across, and driven into, a scene by the blast. The discipline divides cleanly into two questions: before a blast, identifying a suspected bulk explosive (a recovered device, a quantity of material), which is ordinary controlled-substance-style chemistry on an unusual analyte; and after a blast, the far harder problem of post-blast residue.

Explosives residue is the trace chemical material left after an explosion (or left by handling an explosive) — unconsumed explosive, its characteristic breakdown products, and primer or propellant traces — recoverable in microscopic quantities from a blast scene, debris, or a suspect's hands, clothing, or environment. The forensic value is high: identifying the residue can establish what kind of explosive was used (a military high explosive, a commercial blasting agent, an improvised mixture such as ammonium nitrate and fuel oil, a peroxide-based improvised explosive), which in turn constrains the investigation. But the conditions make it treacherous, and the treachery is the lesson.

It helps to know that explosives fall into two broad chemical families, because the families leave different residues and demand different instruments. Low explosives — black powder, smokeless powder, common pyrotechnic mixtures — deflagrate (burn very rapidly) rather than detonate, and they are typically built from an oxidizer and a fuel. Their forensically useful residues are largely inorganic ions: nitrate, nitrite, chlorate, perchlorate, and the metal cations of the mixture. High explosives — military and commercial materials such as TNT, RDX, PETN, and the like, and many improvised organic explosives — detonate, and they tend to be discrete organic molecules whose unconsumed traces and decomposition products are the analytical target. The practical consequence is that explosives-residue analysis is rarely one test: the analyst must cover both the inorganic-ion world (where ion chromatography and elemental methods shine) and the organic-molecule world (where chromatography and mass spectrometry shine), because a single scene may carry either or both. This breadth is one reason explosives casework is specialized — and one reason a single negative result on one method proves so little about the others.

🔬 At the Bench Post-blast work is collection-driven: the analysis is only as good as what comes off the scene, and the scene is chaos. Investigators swab surfaces near the seat of the blast, collect debris and fragments into clean containers, and take control samples from comparable surfaces away from the blast to establish background. In the lab, swabs and extracts are screened (color spot tests for explosive classes exist, analogous to drug color tests, plus presumptive electronic detectors) and then confirmed instrumentally. The confirmatory toolkit (Chapter 23) is broad because explosives are chemically diverse: ion chromatography for the inorganic ions of low explosives and oxidizers (nitrate, nitrite, chlorate, perchlorate), GC-MS and HPLC for organic high explosives, Raman and FTIR for intact particles, and SEM-EDX for the elemental signature of inorganic primer and explosive residues at the particle level. The same funnel — screen, then confirm by an orthogonal instrument — governs here too.

Three problems bound explosives-residue analysis, and an honest analyst names them:

• The sample is minuscule and may be gone. Trace recovery is the norm; a clean scene, weathering, firefighting water, and the blast's own efficiency can leave essentially nothing to find. Absence of detected residue is not evidence that no explosive was used — a vital piece of evidence-strength honesty. You cannot reason from "we found no residue" to "there was no bomb."
• Contamination is everywhere and consequential. Some explosive-associated compounds are environmentally common (nitrates are in fertilizers and soil; certain ions are ubiquitous), and trace analysis is so sensitive that it can detect background that has nothing to do with the crime. Worse, investigators and equipment can carry contamination between scenes. Rigorous control samples and clean-handling protocols are not bureaucratic niceties; they are the difference between a real association and an artifact.
• Interpretation must be conservative. Detecting a compound consistent with an explosive is not the same as proving a particular device, a particular source, or a particular person's involvement. The honest claim is at the level of "residue consistent with explosive type X was detected," with the contamination caveats attached — not "the defendant handled this bomb."

🧠 Cognitive-Bias Watch Explosives cases are high-profile, high-pressure, and emotionally charged — exactly the conditions under which expectation contaminates interpretation (the danger we name fully in Chapter 31). An analyst who knows investigators are confident in a particular suspect, or who is told what residue is "expected," may read an ambiguous trace toward that expectation, or treat a contamination-prone finding as more probative than it is. The safeguard is the same as everywhere in this book: keep domain-irrelevant information away from the bench, insist on controls, and let an independent analyst verify a critical identification blind. The stakes of a terrorism investigation make the bias safeguards more important here, not less.

🔬 Read the Evidence

text FIGURE 21.1 — "A swab from the crater rim" [constructed teaching example] THE ITEM A cotton swab of a steel fragment recovered near the seat of an explosion, submitted with three control swabs: comparable steel from outside the blast zone, an unused swab from the same lot, and a swab of the collecting officer's gloves. THE CONTEXT Collected on day 1, packaged in clean containers, chain of custody intact. Screened by a presumptive spot test for oxidizer ions; the screen was positive for nitrate. WHAT IT SHOWS Instrumental confirmation (ion chromatography, Chapter 23) detects nitrate and a second ion consistent with a specific oxidizer mixture; the three control swabs are clean (no nitrate above background). WHAT IT DOESN'T It does not, by itself, identify who built or placed the device, when the material was deposited, or the exact commercial source. Nitrate is environmentally common, which is why the clean controls — not the positive alone — make the finding meaningful. THE INFERENCE The residue is *consistent with*, and (given clean controls and an orthogonal confirmation) *supports*, the use of an oxidizer-based explosive at this location. A substance-and-place statement, not a person statement. THE LESSON In trace explosives work, the control samples carry as much weight as the positive. A nitrate "hit" with no controls is an artifact waiting to be overstated; the same hit against clean controls and a confirmatory instrument is real evidence — about a substance, never (by itself) about a suspect.

Walk through why that figure turns on the controls. The positive nitrate result, alone, proves almost nothing — nitrate is in fertilizer, soil, sweat, and a hundred ordinary places, and a sufficiently sensitive method finds it everywhere. What converts the result from noise into evidence is the comparison: the blast-zone swab carries the oxidizer signature, and the matched control swabs — from outside the zone, from the unused swab lot, from the officer's gloves — do not. That contrast is what lets the analyst say the residue is associated with the event rather than with the background or with contamination introduced during collection. The instrument confirms the identity of the compound; the controls establish that its presence means something. Strip either away and the conclusion collapses. This is the explosives version of a principle that recurs across forensic chemistry: a detection is only as good as the controls that give it context.

21.5 Fire-debris chemistry (preview Ch. 22–23)

Now to the family of chemistry that carries the cold case and bridges to the next two chapters. When a fire is suspected of being deliberately set, one question for the laboratory is whether the debris carries the chemical residue of a substance used to start or accelerate it. That residue has a name we are previewing here and that Chapter 22 will own and define in full — an ignitable-liquid residue, the laboratory-detectable remnant of a flammable liquid such as gasoline, kerosene, or charcoal lighter fluid recovered from fire debris. The chemistry of detecting it is squarely this chapter's business, and it is, once again, the two-stage funnel — with a special wrinkle that makes the presumptive/confirmatory distinction matter more here than almost anywhere.

The wrinkle is this: for most of the twentieth century, fire investigation tried to identify accelerants by eye, reading visual burn patterns on a floor as if they were chemical proof. Chapter 5 introduced you to where that led — to the execution of Cameron Todd Willingham on the strength of "pour patterns," "crazed glass," and low burn marks that the investigators of the day treated as the unmistakable signatures of a liquid accelerant. Modern fire science (Chapter 22) demolished that folklore: those visual "indicators" are routinely produced by ordinary fire behavior — above all by flashover, when a room's contents all ignite at once — with no accelerant present at all. The lesson for chemistry is precise and central to this chapter: a burn pattern is not a chemical test. The only defensible way to say an accelerant was present is to detect its chemistry in the debris, by the funnel below — never to infer it from the shape of the char. Willingham's case is, among other things, the catastrophe that follows when a visual presumption is mistaken for a chemical confirmation. We advance his case here from exactly that angle; Chapter 22 will take up the fire dynamics in full.

So how is the chemistry actually done? The funnel, adapted to fire debris:

   FIRE-DEBRIS CHEMISTRY — THE FUNNEL, AND WHERE EACH STAGE'S CLAIM ENDS
   (the cold case's gasoline finding lives at the PRESUMPTIVE stage)

   FIRE SCENE
     │  collect charred debris into a CLEAN, UNUSED metal can or nylon bag; seal;
     │  take a CONTROL sample of unburned material; chain of custody
     ▼
   [ FIELD / PRESUMPTIVE INDICATION ]
     │  • investigator's trained nose ("smells of gasoline")
     │  • a hydrocarbon "sniffer" / handheld detector at the scene
     │  • a trained accelerant-detection canine alerting to the debris
     │       → these SCREEN. They INDICATE. They do not identify.
     │       honest status: "accelerant INDICATED"
     ▼
   [ LAB EXTRACTION + CONFIRMATORY INSTRUMENT ]  (Chapter 23)
     │  passive headspace concentration onto an adsorbent, then GC-MS
     │       → separates and identifies the hydrocarbon pattern of the liquid
     │       honest status: "gasoline IDENTIFIED" (a recognized chromatographic pattern)
     ▼
   CLASSIFY the ignitable liquid per the standard scheme (gasoline, petroleum
   distillate, etc.) — an instrumental, defensible identification

The collection step deserves a sentence of its own, because it is where the chemistry is won or lost (and where the CSI track earns its keep). Fire-debris samples are sealed into clean, unused metal cans (the kind paint comes in) or special nylon bags, never ordinary plastic, because common plastics can off-gas hydrocarbons that contaminate the sample or can let the volatile residue escape. A control sample of comparable unburned material is collected so the lab can distinguish a genuine accelerant from the background hydrocarbons that ordinary building materials, carpet, and furnishings release when they burn — a real and serious confound, since many household materials produce pyrolysis products that can resemble parts of an ignitable liquid. Get the collection wrong and even a perfect instrument cannot recover the answer.

Now the funnel's two stages, sharply distinguished, because the cold case sits precisely at the boundary:

• The presumptive/field stage — the investigator's nose, a handheld hydrocarbon detector, or a trained accelerant-detection canine — screens. A dog's alert, in particular, is a sensitive and useful field indicator, and a good one can detect minute quantities. But like every screen in this chapter, it indicates; it does not identify. Dogs can alert to background pyrolysis products, and an alert unconfirmed by the laboratory is a lead, not a result. The honest status after the field stage is accelerant indicated.
• The confirmatory stage — laboratory extraction (typically passive headspace concentration) followed by gas chromatography–mass spectrometry (GC-MS), which Chapter 23 owns — identifies. GC-MS separates the complex mixture in the debris and matches its pattern against the known chromatographic "fingerprint" of gasoline or another ignitable liquid, against published classification standards. The honest status after the lab stage is gasoline identified.

This two-stage structure is not pedantry; it is the exact safeguard whose absence killed Willingham. A field indication — a smell, a sniffer, a dog — is the fire-debris analog of a drug color test: a reason to do confirmatory chemistry, never a substitute for it. Hold that distinction; the cold case is about to land squarely on it.

⚖️ In the Courtroom An expert may testify that an ignitable liquid was identified in fire debris only on the strength of the instrumental result, reported against the recognized classification scheme — not on the strength of a burn pattern, a smell, or a canine alert alone. The defensible sentence is "laboratory analysis identified gasoline in the debris from sample 4." The indefensible ones are "the burn pattern shows an accelerant was poured" (the Willingham error — a visual claim masquerading as chemistry) and "the dog alerted, so gasoline was present" (a screen reported as a confirmation). A cross-examiner who knows this chapter asks two questions: was the field indication confirmed by an instrument? and what control sample ruled out ordinary background hydrocarbons?

21.6 The limits of presumptive tests: the roadside "field test" problem

We end where the chapter's stakes are highest and most human. Everything above has insisted, in the abstract, that a presumptive test is a screen and not an identification. The roadside drug field test is where that abstraction becomes a person in a jail cell — and where the failure to honor it has produced documented wrongful arrests and wrongful convictions across the United States.

The roadside field test is a small, cheap kit: a pouch or vial of reagent into which an officer drops a bit of a suspected substance during a stop. It is a presumptive color test — the same cobalt-thiocyanate and Marquis-type chemistry from §21.2, packaged for the field. And it carries every limitation of a color test, now operating in the worst possible conditions for reliability: in variable light and temperature, read by an officer who is not a trained chemist, with no controls, on a substance whose identity is unknown, with someone's liberty riding on a color the officer believes they see. The kits are known to produce false positives on a startling range of ordinary, legal substances, and the reading of an ambiguous color by a non-specialist under roadside conditions adds a further, well-documented layer of error.

Here is the chain of harm, and it is not hypothetical. A field test "turns positive." The person is arrested and booked. Facing a felony drug charge, a setting in which bail may be unaffordable and a trial months away, many defendants accept a quick plea bargain — pleading guilty to a lesser charge to go home — before the substance is ever confirmed by a laboratory. The presumptive screen, which should have been only the first step of a two-stage process, becomes in practice the last step: the conviction is entered on the color test alone, and the confirmatory instrumental analysis that would have caught the error is never run, or is run only later, if ever. When crime laboratories have gone back and tested the substances in such cases, they have found, in documented instances, that the "drug" was no drug at all.

⚠️ Junk-Science Alert The roadside field test is not junk chemistry — a color test is a legitimate screen. It becomes a junk-science engine through misuse: a presumptive result treated as a confirmatory identification, with the confirmatory stage skipped entirely. This is the purest possible illustration of the chapter's spine. The method is fine in its lane (triage, probable cause). The catastrophe is structural — a system that lets a screen convict. The fix is not a better color; it is the discipline, and the policy, to never let a presumptive result stand as an identification.

This is a textbook case of the CSI effect cutting both ways (Theme 4) joined to the not-all-methods-are-equal theme (Theme 2). The cultural faith that a "test" delivers a verdict — that a kit beeping or a vial turning blue is the answer, the way it always is on television — collapses the distinction the whole field depends on. A juror, a prosecutor, or a defendant who understood that a color test only suggests, and that identification requires an orthogonal instrument, would not accept a field test as proof. The remedy is the one this entire chapter has been building toward: confirmatory analysis before conviction, every time, without exception, because the screen was never designed to bear that weight.

Where do these methods sit on the validity spectrum? Precisely: presumptive color and microcrystalline tests are valid screens and invalid identifications. As triage — directing the lab, supporting a seizure or a probable-cause arrest, reliably excluding a class when negative — they are sound and have their place. As the sole basis for identifying a substance, they have no business on the strong end of the spectrum where confirmed instrumental chemistry lives; used that way, they are an overstatement of exactly the kind the NAS 2009 and PCAST 2016 reports warned against. Instrumental confirmation (Chapter 23) is what earns the substance-identification its place near DNA. The screen, alone, does not — and a system that pretends it does manufactures the wrongful convictions this book exists to prevent.

🔍 Check Your Understanding 1. Explain, in one sentence each, the two distinct points in the "chain of harm" where confirmatory testing would have prevented a wrongful drug conviction. 2. A roadside kit is "a valid screen and an invalid identification." Give one legitimate use of the kit and one illegitimate use, and name the principle that separates them.

🗂️ The Case File

The fire debris from Mill Creek. By this point in the investigation, the assumption of an accidental fire is already gone: the autopsy established that Marcus Diallo was dead before the fire (Chapter 11), the trauma was real and not a heat artifact (Chapter 12), and toxicology has shown a sedative at an incapacitating level in his blood (Chapter 20). The question the fire now raises is not whether he died in an accident — he did not — but how the fire itself was started, and whether someone used an accelerant to set it. That question routes to the chemistry bench, and this chapter is where its first answer arrives.

When the scene was worked, debris was collected from several areas of the cabin's front room — charred flooring and subfloor — into clean, unused metal cans, with a control sample of comparable unburned material, chain of custody intact (the collection done correctly, in contrast to the first-responder errors of Chapter 2). At the scene, two field indications pointed the same way: the responding investigators noted a hydrocarbon odor consistent with gasoline, and a handheld hydrocarbon detector indicated volatiles in the front-room debris. On that basis, the field finding is recorded at its true, careful strength: accelerant indicated — consistent with gasoline.

State the strength exactly, because this is the chapter's whole lesson. What we have is a presumptive indication. A smell and a sniffer are screens; they suggest, they do not identify. The honest status is indicated, not confirmed. We have not identified gasoline — that is an instrumental claim, and the GC-MS confirmation belongs to Chapter 23, which will either confirm the field indication against the recognized chromatographic pattern of gasoline or fail to. And note what this chapter explicitly does not let us do: we cannot call the fire "arson" from the smell of gasoline, because (as Chapter 22 will establish on valid grounds) an incendiary finding requires fire-science analysis of origin and cause, not a single accelerant indication — and certainly not the discredited burn-pattern folklore that killed Willingham (§21.5). The chemistry indicates an accelerant; the arson conclusion is Chapter 22's to earn.

Running status. No person is included or excluded by this evidence. What changes is the picture of the fire: an accelerant is now indicated in the debris, which — combined with the pre-fire death and the sedative — makes a deliberately set, staged fire a live hypothesis rather than an assumption to be disproved. Log it in the workbook (Appendix I) as a presumptive finding, status accelerant indicated — gasoline, with the confirmation flagged as pending (Chapter 23). Resist the pull to write "gasoline confirmed" or "arson"; the science has earned neither word yet. It is a capital mistake to theorize before one has data — and what we have, so far, is an indication awaiting an instrument.

Conclusion

Forensic chemistry is, on its merits, one of the strongest disciplines in this book — its core claim, "this substance is X," rests on validated analytical chemistry, carries measurable error rates, and ends in an instrumental result that can be checked and defended. But that strength belongs to the end of a two-stage process, and the chapter's single lesson is to never confuse the stages. A presumptive test — a color in a tube, a crystal under glass, a kit at the roadside, a sniffer or a dog at a fire scene — screens: it is fast, cheap, sensitive, and genuinely useful for triage and for reliably excluding a class, but it indicates consistency with a class and can be fooled, and it is not an identification. A confirmatory test — the orthogonal instruments of Chapter 23 — identifies: it interrogates the molecule by an independent physical principle, and it is what may honestly be reported as "is" and defended on the stand.

We watched that distinction carry real weight in three domains. In drug chemistry, the failure to confirm has put innocent people in jail on a color a roadside officer believed they saw — the purest demonstration in the book that a screen treated as a verdict is a junk-science engine, however sound the screen is in its lane. In explosives work, where the sample is often nearly gone, we saw that a detection means nothing without the control samples that give it context, and that absence of residue is not absence of a bomb. And in fire debris, we set up the arson reckoning of Chapter 22 and the instruments of Chapter 23, and advanced the Willingham catastrophe from the chemistry angle: a burn pattern is not a chemical test, and the only defensible accelerant finding is a detected one. The cold case landed exactly on that boundary — an accelerant indicated, gasoline not yet confirmed — and the honesty of that "indicated, not confirmed" is the chapter in miniature.

Chapter 22 takes up the fire itself — origin and cause, flashover, the debunked indicators, and how an incendiary finding is built on valid grounds and explicitly contrasted with the folklore that killed Willingham. Chapter 23 then brings the instruments that turn this chapter's "indicated" into "confirmed." Carry one sentence into both: the screen suggests; the instrument confirms; and the difference between them is the difference between an honest forensic science and a wrongful conviction.

Key Terms

• Forensic chemistry — the application of analytical chemistry to legal questions, chiefly the identification and characterization of substances such as controlled drugs, explosives and their residues, and fire accelerants.
• Controlled-substance analysis — the forensic identification of illegal drugs and regulated pharmaceuticals (what a seized substance is, and often how much is present), the highest-volume work of the chemistry section.
• Presumptive color test — a chemical screening test in which a reagent produces a characteristic color when it contacts a class of substance, indicating that the class is probably present (or, when negative, probably absent); class-level and fallible, never an identification.
• Microcrystalline test — a presumptive technique in which a reagent forms crystals of a characteristic shape and habit with a target compound on a microscope slide, read by the analyst under magnification; more discriminating than a color test but still presumptive and partly subjective.
• Explosives residue — the trace chemical material left after an explosion or by handling an explosive (unconsumed explosive, characteristic breakdown products, primer/propellant traces), recoverable in microscopic quantities and meaningful only against rigorous control samples.

Spaced Review

1. A drug report reads: "Item 1 is cocaine," but the file shows only a Scott (cobalt thiocyanate) color test was performed. Identify the overstatement and write the defensible report sentence. (§21.2–21.3)
2. Recall the serology funnel from Chapter 10 (presumptive vs. confirmatory). Explain how the drug-chemistry funnel in this chapter is the same logic, and name one way a presumptive negative is the more reliable result in both. (§21.3; Chapter 10)
3. In the cold case, the fire debris yields "accelerant indicated — gasoline," not "gasoline confirmed." Using §21.5, explain precisely what is still missing and which chapter supplies it — and why the smell of gasoline cannot, by itself, make the fire "arson." (§21.5; preview of Chapters 22–23)
4. Validity-spectrum question. Where do presumptive color tests sit on the NAS 2009 / PCAST 2016 spectrum as screens versus as identifications, and what single addition moves a substance identification to the strong end where DNA lives? (§21.6; and the spectrum from Chapter 1)
5. Explosives-residue analysis detected nitrate near a blast, but the matched control swabs were never collected. Using §21.4, explain why the bare positive is nearly worthless and what the missing controls would have established. (§21.4)