INTERPRETATIONS
Statistically evaluating trace evidence, including paint, is difficult. Most of the statements that can be made relate to samples tested in clinical or research trials although some are based on actual casework samples. A consensus of forensic paint examiners is that the following factors strengthen an association between two analytically indistinguishable paint samples:
• the number of layers;
• the sequence of layers;
• the color of each layer;
• cross-transfer of paint between items.
One of the best-known forensic paint examiners in the United States, Scott Ryland of the Florida Department of Law Enforcement forensic laboratory in Orlando and his colleagues have stated that an association between two paint samples with six or more correlating layers indicates that the chance the samples originated from two different sources is “extremely remote.” In cases with evidence this strong, merely stating that the two samples “could have had a common origin” is not enough; that level of state ment undermines the strength of a six-layer-plus association. Although an answer is not statistical or mathematical doesn’t mean the statement isn’t accurate, valid, and sound. The significance of architectural paints varies and is in general not as well documented in the literature. This is most likely due to the enormous variability in colors, application styles, and the application of the paint itself (not all brushstrokes are equal, resulting in highly variable layers between samples). The situation is simi lar with spray paints, about which even less is known.
Instances of generating statistics to assess the evidentiary value of paint have been attempted in both the clinical literature and in casework. They are based, as are most manufacturing inquiries, on the concept of a batch lot, a unit of production and sampling that contains a set of analytically indistinguishable products. For example, a batch tank of automotive paint of a given color may hold 500 to 10,000 gallons, which would color between 170 and 1600 vehicles. This would then be the unit of comparison for the significance of an automotive paint com parison—the manufacturing batch lot. If analytically identifiable differences can be determined between batch lots, then the base population is set for any other analytically indistinguishable paint samples. The final significance will be determined by the number of vehicles in the area at the time of the crime and other characteristics that set that sample apart (very rare or very common makes or models). By comparison, a batch lot of architectural paint may be from 100 to 4500 gallons.
As with other forensic sciences, a reference collection is essential for train ing and casework, and paint is no exception. For paint, such collections take two forms: documented samples and data. From 1974 until 1989, the National Bureau of Standards (now the National Institute of Standards and Technology) published a reference collection of automotive paint colors derived from actual production samples. Later in the collection’s history, chemically accurate samples were added to it. Although limited in time frame and representation (it contained only US-manufactured colors), this type of collection is invaluable for training and make and model searches. The FBI Laboratory Division maintains a color and chemical reference collection that is housed at the new laboratory in Quantico, Virginia; it is not a lending library, however, and the samples must remain at the laboratory.
A different approach overcomes these limitations. The Paint Data Query (PDQ) project is run by the Royal Canadian Mounted Police (RCMP) and is the largest international automotive paint database. It contains nearly 20,000 samples of paint systems, which represent over 74,000 individual paint layers used on most domestic and foreign vehicles being sold in North America, Australia, and Asia. To use the PDQ database, forensic scientists enter the color, chemical composition, and layer sequence information taken from unknown paint chip(s) recovered from a crime scene. Forensic laboratories can obtain the data by participating in PDQ by submitting 50 automotive paint samples per year, with some rare entries earning “double points.” The database contains information on layer structures, primer colors, binders, pigment chemistry, and topcoat chemistry, in both visual and spectrometric formats.
