Background
The Massachusetts Supreme Judicial Court vacated Keith Williams’s first-degree murder conviction and three related convictions because prosecutors used scientifically negative gunshot-residue tests as affirmative proof that he was the shooter. Justice Elizabeth N. Dewar wrote for the court. The ruling turns on a practical evidentiary limit: when the Commonwealth’s own expert says a laboratory result supports no inference, a prosecutor cannot invite jurors to treat details beneath the laboratory’s validated threshold as identification evidence without first establishing that use as reliable.
The charges arose from a 2014 shooting during a parade on Blue Hill Avenue in Dorchester. Gunfire aimed toward two men instead killed bystander Dawn Jaffier and wounded Lealah Fulton. No witness identified Williams as the shooter, but witnesses placed him with two companions shortly afterward, saw him holding and disposing of the revolver, and described the shooter in ways that partly matched his build. A jury convicted him in 2017 of first-degree murder, armed assault with intent to murder, assault and battery by means of a dangerous weapon, and carrying a firearm without a license.
Investigators tested Williams and his two companions for gunshot residue. Under the FBI and Massachusetts State Police protocol described at trial, a positive result required three qualifying particles in a single sample. Every sample was negative. One particle nevertheless appeared on a sample from Williams’s hand and another on his T-shirt; none appeared on his companions’ samples. The Commonwealth emphasized that difference in its opening and closing, even though its expert testified that “nothing” could be drawn from a negative result.
The Court’s Holding
The SJC held that the subthreshold particle evidence should not have been admitted to show that Williams, rather than either companion, fired the gun. Massachusetts evidence law permits relevant expert evidence only when the underlying theory or methodology is reliable for the purpose for which it is offered. The court did not announce that below-threshold scientific observations are categorically inadmissible. Instead, it focused on the Commonwealth’s failure to establish that one particle in each of two negative tests reliably supported the identification inference urged at trial.
The expert’s testimony defeated, rather than supplied, that foundation. He explained the three-particle threshold, confirmed that separate samples could not be combined to reach it, and said no conclusion could be drawn from any negative test. The prosecutor nonetheless told jurors that the laboratory protocol was not “a jury standard of practice,” invited them to interpret the evidence as they wished, and argued that particles found only in Williams’s samples distinguished him from the other men. Without evidence validating that comparison, the inference was unsupported.
The error was prejudicial because identification was contested and the remaining proof was not overwhelming. Although Williams was seen with the weapon moments after the shooting and evidence supported consciousness of guilt, no eyewitness or forensic test positively identified him as the shooter, and some descriptions pointed toward another man. The court therefore ordered a new trial on the murder, armed-assault, and dangerous-weapon counts. It also vacated the firearm-license conviction because the jury were not instructed that the Commonwealth had to prove Williams lacked a valid license. The SJC rejected Williams’s sufficiency challenge, however: repeatedly firing after another man pointed a gun permitted a rational finding of deliberate premeditation, so a first-degree murder retrial is allowed.
Key Takeaways
- A negative forensic test cannot be converted into affirmative identification evidence merely by highlighting measurements that fall below the validated positive threshold.
- The proponent must establish that a scientific observation is reliable for the precise inference presented to the jury; general reliability of the testing technology is not enough.
- Prejudice is especially likely when prosecutors feature unsupported forensic evidence in opening and closing and the remaining identification evidence is disputed.
- Vacatur for trial error does not bar retrial where the properly admitted evidence was sufficient, and deliberate premeditation may be inferred from a sequence lasting only seconds.
Why It Matters
Williams gives Massachusetts trial lawyers a focused framework for disputes over low-level or inconclusive forensic findings. The important question is not simply whether gunshot-residue analysis, DNA testing, toxicology, or another discipline is generally accepted. Counsel and judges must ask whether the evidence reliably supports the particular comparison or conclusion the jury is being asked to make. A result labeled negative cannot carry a positive implication without a demonstrated scientific basis.
The opinion also warns against using closing argument to outrun an expert’s limiting testimony. Prosecutors may not tell jurors that they can disregard a laboratory threshold when the record contains no evidence validating an alternative interpretation. Defense counsel, meanwhile, should object both to admission and to argument, press experts on what a negative or inconclusive result actually means, and separate scientific limits from ordinary factual inferences. For retrials and future prosecutions, the Commonwealth may offer only uses of subthreshold data that it can support with an adequate reliability foundation.