Paint Databases (PDQ, EUCAP) and Comparison Casework

PDQ was created in 1976 at the RCMP Forensic Laboratory Services in Ottawa, initially as a paper record of automotive OEM paint formulations voluntarily submitted by vehicle manufacturers. The transition to a relational database began in the 1980s; the system became web-accessible to participating forensic laboratories in the late 1990s. Access is restricted to subscribing forensic science institutions and law enforcement agencies; as of 2024, subscribers include forensic laboratories in the US (FBI, ATF, state crime labs), Canada (RCMP and provincial labs), the UK (former Forensic Science Service, now commercial providers), Australia (AFP, state laboratories), and, since 2015, selected laboratories in India and Southeast Asia.

What PDQ contains. Each database entry corresponds to a specific vehicle manufacturer, model, model year, paint colour code, and assembly plant. The entry records the physical description of each paint layer (number of layers, colour, approximate thickness), the FTIR spectrum of each layer (binder fingerprint), and the Raman spectrum of the basecoat pigments where available. More recent entries include Py-GC-MS pyrograms for the key layers and SEM-EDS elemental profiles for the primer. Manufacturer entries are submitted directly by the OEM companies as part of the PDQ cooperation agreement; refinish-product entries are submitted by the major refinish-product manufacturers (BASF Glasurit, PPG, Axalta) and cover their product lines' spectral fingerprints.

Querying PDQ. A forensic analyst queries PDQ by entering the physical and spectroscopic data from the questioned chip: layer count, layer colours, FTIR spectra (uploaded as digital spectral files or entered as peak tables), Raman pigment identification, and any SEM-EDS elemental data. The database returns a ranked list of candidate formulations ordered by similarity score. The similarity scoring uses a combination of spectral cross-correlation (for FTIR profile matching), peak position matching (for Raman peaks), and layer-sequence comparison. A score of 0.90 or above on the FTIR correlation is treated as a strong candidate match; scores below 0.70 are typically not pursued further. The analyst reviews the top-ranked candidates, downloads the reference spectra, and performs a visual overlay comparison before accepting a candidate match.

PDQ output and its evidential meaning. A PDQ match narrows the field of possible source vehicles to those manufactured in specific model years, by specific manufacturers, in specific colour codes. It does not identify a specific vehicle; for that, a direct comparison between the questioned chip and a physical sample from the suspect vehicle is necessary. The PDQ match is typically reported as "the paint is consistent with the PDQ entry for [manufacturer] [model] [model year range] in colour code [code]." This language supports the investigating officer's direction of enquiry (seize vehicles matching the make/model/year/colour) and provides the starting point for a direct comparison if a suspect vehicle is located.

PDQ in US casework. US forensic laboratories use PDQ within the SWGMAT examination framework. The FBI Paint Unit at Quantico maintains its own curated automotive paint reference collection alongside PDQ subscription, and FBI paint examiners testify in federal courts as qualified experts in automotive paint comparison. Under the Daubert standard, PDQ-supported testimony has been admitted in numerous federal and state courts because PDQ represents a peer-reviewed, validated database maintained by an independent national forensic laboratory. The critical foundational requirement is that the analyst document the PDQ match criteria and perform a direct physical comparison, not merely report a database match in isolation.

EUCAP, the European Collection of Automotive Paints, is maintained at the Bundeskriminalamt (BKA) in Wiesbaden, Germany. It was established in the 1990s as a collaborative project of the ENFSI Expert Working Group on Paint (EPG) and holds formulations from European vehicle manufacturers, particularly the Volkswagen Group (Volkswagen, Audi, Porsche, Seat, Skoda), BMW Group, Daimler (Mercedes-Benz), PSA Group (Peugeot, Citroën), Renault, and Fiat. It also includes entries from Japanese and Korean manufacturers for models sold in the European market.

How EUCAP differs from PDQ. The two databases are complementary rather than competing. PDQ has greater depth for North American market vehicles and for Japanese brands sold in North America; EUCAP has greater depth for European market vehicles and for refinish products in the European market. The ENFSI EPG protocol recommends querying both databases when the vehicle make is unknown, because a formulation poorly represented in PDQ may have a complete record in EUCAP. The BKA processes several thousand paint-examination requests per year from German federal and state police, and the EUCAP database supports the majority of those examinations.

Data format and spectral quality. EUCAP entries include FTIR spectra for each layer (collected under standardised ATR conditions), colour measurements (CIE Lab* values), and layer-thickness measurements. More recent additions include Raman spectra and SEM-EDS elemental profiles. The standardisation of ATR collection conditions (a specific crystal material, contact pressure, and number of scans) allows spectra from different contributing laboratories to be compared with calibration-adjusted confidence, a logistical improvement over the more variable collection modes represented in older PDQ entries.

EUCAP in European casework. Across the EU, paint comparison in hit-and-run casework is governed by the ENFSI EPG guidelines, which treat a EUCAP match as a candidate identification requiring direct comparison with a reference sample from a suspect vehicle before a conclusion can be stated. German Landeskriminalamt (LKA) laboratories, which handle the volume of German hit-and-run paint casework, query EUCAP as the first-line database and PDQ as a secondary source. UK commercial forensic providers (LGC, Axiom International) access both databases. The FSR Codes of Practice require that any database query be documented in the examination case file, including the search parameters, the candidate list, and the basis for selecting or rejecting candidates.

The forensic paint comparison protocol is formalised in two parallel documents: SWGMAT's Forensic Paint Analysis and Comparison Guidelines (US) and the ENFSI EPG protocols (Europe). Both follow the same four-stage logic: physical characterisation, microscopic examination, chemical analysis, database query and direct comparison.

Stage 1 and 2: Physical and microscopic. Cross-section layer count, colour, and thickness establish whether the questioned chip is physically consistent with the database candidate (e.g., a candidate specifying four layers, silver metallic basecoat, and approximate layer thickness range of 15-25 µm for the clearcoat). Physical inconsistency at this stage eliminates the candidate and redirects the query.

Stage 3: Chemical comparison. FTIR spectra of each layer in the questioned chip are compared overlay-on-overlay with the corresponding layer spectra from the PDQ or EUCAP entry. A high spectral correlation, defined as no significant unexplained peak differences above the background noise level, supports inclusion. Any spectral peak in the questioned chip that is absent from the reference entry, or vice versa, is a potential exclusion if it cannot be explained by instrumental variation or sample history (weathering, contamination).

Stage 4: Direct comparison with a reference sample. After a PDQ or EUCAP match identifies a candidate vehicle, and if a suspect vehicle is located, the analyst collects a physical paint sample from the suspect vehicle (typically from an undamaged panel near the damage area, to avoid body-shop repair contamination) and prepares a cross-section for direct comparison alongside the questioned chip. This is where the comparison microscope (side-by-side optical bridge comparison, as described in Module 2 and Module 8) and the full spectroscopic sequence are repeated on corresponding layers. The direct comparison is always the primary evidence; the database match is the supporting context.

Reporting the conclusion. Both SWGMAT and ENFSI guidelines provide a categorical conclusion scale. The SWGMAT five-point scale runs from "excluded as originating from the same source" through "could have originated from" and "consistent with having originated from" to "associated with" and (at the highest evidential level) "is indistinguishable from the sample from the suspected source." The ENFSI scale is similar but uses a verbal likelihood-ratio framing: "strong support for a common source," "moderate support," "inconclusive," "moderate support against a common source," and "strong support against a common source." Neither system uses the word "match" as a standalone conclusion, because it implies certainty that the evidence cannot provide.

An increasing number of forensic laboratories in the Netherlands (NFI), Finland, the UK, and Australia express paint comparison conclusions as numerical likelihood ratios or as verbal expressions calibrated to a stated LR range. This approach is consistent with the ENFSI FIRM (Forensic Intelligence and Reporting Model) and with recommendations from the NIST-hosted OSAC Forensic Units Standards Committee. It is not yet standard practice in North American or Indian casework, but the trend is toward more formal probabilistic reporting in all trace-evidence disciplines.

Automotive paint evidence in hit-and-run investigations follows a consistent workflow internationally, but the legal context, evidentiary standards, and institutional resources differ significantly between jurisdictions.

United States: the Christopher Coleman case and FBI paint evidence. The 2009 murders of Sheri, Gavin, and Garret Coleman in Columbia, Illinois are primarily known as a domestic homicide case rather than a traffic matter, but the paint evidence workflow that arose in related investigative work is representative of FBI-handled automotive paint comparison. More directly relevant to hit-and-run is the body of FBI paint-evidence work in vehicular homicide investigations across federal and state courts. The FBI Paint Unit has processed paint evidence in thousands of US cases since the 1980s. Under the Daubert gatekeeping framework, FBI paint examiners testify that the PDQ-supported comparison methodology is empirically tested, peer-reviewed in publications including the Journal of Forensic Sciences and Forensic Science International, and generally accepted by the forensic community. The critical admissibility requirement post-Daubert is that the error rate for false associations be stated; a 1998 study by the FBI Paint Unit, published in the Journal of Forensic Sciences, estimated the probability of two different OEM automotive paint systems producing indistinguishable four-layer FTIR/Raman profiles at less than 1 in 10,000 for same-colour comparisons, a figure cited in court as an indicator of discrimination power.

United Kingdom: R v. Adams and paint evidence in Crown Court. The Court of Appeal decision in R v. Adams (1996 EWCA Crim) addressed the admissibility and jury presentation of DNA statistical evidence and Bayesian reasoning, and the principles it established have been considered in subsequent expert-evidence cases including those involving trace evidence. In UK Crown Court practice, the Forensic Science Regulator's Codes of Practice (FSR-C-100, updated 2020) require that all trace-evidence examinations be performed by ISO 17025-accredited laboratories. Commercial providers including LGC Forensics and Cellmark handle automotive paint comparisons for police services and the Crown Prosecution Service. The UK court expert report must comply with Criminal Procedure Rules Part 19, which requires the expert to identify any range of opinion on the matter, the basis for their opinion within that range, and any uncertainties. UK paint examiners rarely use numerical LR values in court reports at present, but the FSR strategy documents (including the Cognitive Bias Mitigation and Reporting Strategy 2019) push toward more quantitative uncertainty expression.

European Union: ENFSI EPG and cross-border cooperation. Within the EU, the ENFSI EPG protocols create a methodological common ground for paint comparison across member states. Cross-border hit-and-run cases, particularly involving commercial vehicles, are handled through Eurojust and Europol evidence-sharing mechanisms. EUCAP database queries can be performed by forensic laboratories in any ENFSI member country, allowing a fragment collected in, say, Austria to be queried against formulations for vehicles manufactured and sold primarily in Germany or France. The BKA acts as the EUCAP custodian and processes inter-laboratory queries on request.

India: BNS 2023 § 106 and hit-and-run casework. In India, culpable homicide or causing death by negligence in road traffic incidents is charged under BNS 2023 § 106 (formerly IPC § 304A). The Motor Vehicles (Amendment) Act 2019 created the hit-and-run scheme under the Motor Vehicles Act 1988, which provides compensation to victims even when the offending vehicle is unidentified, but also increased the incentive for police to identify the vehicle. Paint analysis in Indian hit-and-run casework is conducted by the CFSL, the DFSS, and state forensic science laboratories. Not all state FSLs have full Py-GC-MS capability; some rely on FTIR and comparison microscopy alone. PDQ access was expanded to selected Indian FSLs through an RCMP-DFSS cooperative agreement formalised around 2015-2018, though coverage of Indian-manufactured vehicles (Maruti Suzuki, Tata Motors, Mahindra, Hyundai India) in PDQ is less complete than for North American or European OEMs. The absence of a PDQ match for an Indian-market vehicle does not exclude paint association; a direct comparison against physical samples remains the primary evidence.

Australia and Canada. In Australia, the Australian Federal Police (AFP) and state forensic laboratories handle automotive paint comparison under ANZFSS guidelines that closely follow ENFSI EPG methodology. The AFP paint unit contributed to the development of the global PDQ expansion and has processed casework from hit-and-run incidents involving trans-Pacific vehicle models. In Canada, the RCMP acts as both the custodian of PDQ and as the primary paint examination laboratory for federal casework; provincial police services typically refer paint casework to the RCMP lab or to qualified provincial forensic science services.

Categorical verbal conclusions such as "consistent with a common source" communicate the direction of the analyst's opinion but do not allow the factfinder to quantify the strength of that opinion relative to the alternative. The Bayesian likelihood ratio (LR) framework addresses this by expressing the comparison conclusion as a number: the probability of the observed physical and chemical data given the hypothesis that the samples share a common source (H1), divided by the probability of the same data given the hypothesis that the samples come from different sources (H2). An LR greater than 1 supports H1; an LR much greater than 1 provides strong support.

Calculating LR for paint. The LR for a paint comparison requires an estimate of the frequency of the particular paint profile in the population of potentially relevant vehicles. This frequency is estimated from the PDQ or EUCAP database: how many distinct formulations produce spectra within the acceptance range of the questioned sample? A query that returns only one PDQ candidate out of the entire database implies a high LR for a physical match. A query that returns 50 candidates implies a lower LR. The LR calculation is not routinely performed in all jurisdictions, but the NFI (Netherlands Forensic Institute) and the UK's FSR have published guidance on its application to paint.

The "match probability" concept. In US federal court testimony, FBI paint examiners have used a semi-quantitative "match probability" framing: the probability of randomly selecting, from the population of all vehicles consistent with the scene (same make, model, year, colour), a vehicle whose paint would produce an indistinguishable comparison result. This framing is not identical to a Bayesian LR but communicates similar information to a lay jury. The numerical estimate typically relies on the PDQ hit-rate for a given profile: if only 3 formulations in the 75,000-entry database match the questioned chip's full profile, the "match probability" for a randomly selected vehicle of that colour is approximately 3/75,000. This framing has been criticised by defence experts who argue that the database is not a random sample of all vehicles ever manufactured.

ENFSI FIRM and the verbal LR scale. The ENFSI Forensic Intelligence and Reporting Model proposes a verbal scale tied to LR ranges: "inconclusive" (LR near 1), "limited support" (LR 10-100), "moderate support" (LR 100-10,000), "strong support" (LR 10,000-1,000,000), and "very strong support" (LR above 1,000,000). Several Dutch and Finnish forensic laboratories have adopted this scale for paint-comparison reports in court proceedings. The scale is also used in the ENFSI EPG guidelines as the recommended reporting framework for member laboratories, though implementation varies widely.

India's evidentiary framework. The BSA 2023 does not specify a reporting format for trace-evidence expert opinions, and Indian courts have not adopted a formal LR framework for paint comparison. Expert witnesses in Indian sessions courts typically present their conclusions in categorical terms consistent with the NAS 2009 report recommendations adapted for local practice. The Bharatiya Sakshya Adhiniyam § 23 requires that the court be satisfied that the expert's opinion is relevant and based on knowledge in the relevant field; this is a relevance test, not a methodology-validity test. However, the Supreme Court's emphasis on scientific rigor in technical expert evidence, expressed in cases including State of Himachal Pradesh v. Jai Lal (1999) and subsequent High Court decisions, means that an examiner who can articulate a probability-based opinion under cross-examination is in a stronger position than one who relies purely on categorical language.

Experienced defence attorneys and prosecution counsel in the US, UK, and Australia have developed a standard set of cross-examination vectors for paint-comparison evidence. Forensic scientists can address each of these vulnerabilities in the examination report, reducing exposure during cross-examination.

Failure to perform a direct comparison. The most common and most easily attacked weakness is a report that relies entirely on a PDQ or EUCAP database match without a direct comparison between the questioned chip and a physical sample from the suspect vehicle. A database match identifies a formulation class; it does not identify a specific vehicle. If the suspect vehicle's paint was not sampled, any comparison opinion is limited to "consistent with the make/model/year range" and cannot support identification of a specific vehicle. Defence counsel routinely asks: "Did you compare the chip against a sample from this specific vehicle?" A "no" answer significantly reduces the weight of the opinion.

Layer-sequence contamination during preparation. If the cross-section preparation was performed carelessly, and layers from adjacent chips were mixed during grinding, the resulting layer profile does not represent either chip accurately. The analyst should document the preparation process photographically and retain the unused portion of the chip. SWGMAT guidelines require that the original chip be retained where possible.

Transfer mechanism assumption. A paint comparison establishes that the questioned chip is physically and chemically consistent with the known source, but it cannot by itself establish the mechanism of transfer. Defence counsel may argue that the transfer occurred through a secondary route (e.g., both vehicles were repainted at the same body shop using the same refinish product). The analyst can address this by characterising the layer architecture: a chip retaining the OEM four-layer stack, including the e-coat, came from a vehicle at factory condition, not a body-shop application.

Database coverage gap for non-enrolled manufacturers. PDQ covers primarily North American, European, Japanese, and Korean OEMs from 1976 forward. Vehicles from manufacturers not enrolled in PDQ, including some Chinese brands now entering international markets, and Indian-manufactured vehicles in some model lines, will not produce a database hit. The absence of a PDQ hit should be reported as a negative finding, not as an absence of association, and the examined chip should be fully characterised spectroscopically for direct comparison purposes.

Single-method conclusions. A paint comparison opinion based on FTIR alone, without Raman, SEM-EDS, or Py-GC-MS data, is more vulnerable to cross-examination on discrimination power. SWGMAT guidelines specify that the full suite of methods should be applied, or the reason for omitting a method should be documented. An opinion based on four independent and consistent data streams (physical, FTIR, Raman, elemental) is substantially more robust than one based on visual comparison alone.

Automotive hit-and-run represents the highest volume of forensic paint casework globally, but paint evidence also arises in burglary, art investigation, and industrial incident contexts, where the same comparison methodology applies.

Burglary and tool-mark evidence. Paint transferred from door frames, window sills, and lock casings to a burglar's tools, clothing, or skin is among the most common non-automotive paint evidence class. Architectural latex and alkyd paint comparison in burglary cases typically involves direct comparison (no database equivalent to PDQ for architectural paints), relying on FTIR binder fingerprinting, Raman pigment identification, and SEM-EDS elemental profiling of extenders. The discrimination power is lower than for automotive OEM paints because architectural formulations are less chemically specific. Nevertheless, a combination of binder type (latex versus alkyd), colour (CIE Lab* colorimetry), and extender profile can produce a meaningful associative finding.

Art theft and forgery. Paint from works of fine art is examined using the pigment and binder libraries developed for the art-investigation community: the IRUG database, the Getty Conservation Institute pigment library, and the Gettens-Stout Artists' Pigments volumes. In art-theft cases, a paint flake from a suspect's clothing or vehicle can be compared against a reference sample from the damaged work; the specific pigment combination in a historic work (lead white, smalt, ultramarine, Prussian blue in a defined layer sequence) is effectively unique to that work. Forgery detection relies on the same pigment-analysis tools: the presence of a pigment invented after a painting's claimed date (phthalocyanine blue, invented 1928, in a claimed 17th-century work) is a technical forgery indicator.

Industrial incident investigation. Paint from structural components, industrial equipment, and infrastructure transfers to vehicles and workers in collision and maintenance incidents. In these cases, the characteristic inorganic chemistry of industrial protective coatings (zinc-rich epoxy primers, micaceous iron oxide) provides discrimination against common automotive and architectural paint profiles.