By John Hollway

For decades, lawyers, judges, and juries have turned to science to provide objective truths in criminal justice. TV shows like CSI and Bones suggest that forensic analysis can conclusively solve all manner of crimes. But scientific knowledge and expert opinions constantly evolve. When science marches on, how should the justice system respond?

The conviction of Daniel Dougherty is one example. His arson conviction is one of a number being appealed on the basis of advances in fire science that call into question expert testimony presented in good faith at the time of trial.

Dougherty was charged, convicted of arson, and sentenced to death in 2000 for setting a 1985 fire in which his two young sons were killed. (He is now serving life in prison.)

At the time of his trial, the fire marshal testified that the blaze was arson. But by 2012, the same evidence led an expert witness to conclude that there was no scientific basis to determine the fire had been deliberately set.

In particular, heat marks on the floor that were once considered evidence of an accelerant being used are now understood as the effects of "flashover" - the way fire behaves in very hot spaces depleted of oxygen. Similarly, where shattered glass in a room was once considered evidence that an accelerant had created a very hot interior space, it is now understood to be the result of cooling, caused in all likelihood by water from a fire hose.

Given such changes in the experts' scientific interpretation of this evidence, what should be done?

In Dougherty's case, a panel of three Superior Court judges has ordered a new trial. Philadelphia prosecutors are asking for reconsideration of that decision by the full court.

The District Attorney's Office would do well to approach the case with an open mind and use it as an opportunity to address the broader question of how such cases should be handled. Dougherty's case is not unique, and the issue it raises is not limited to arson. Similar advances in scientific understanding of shaken-baby syndrome, for example, are casting doubt on the scientific basis of a number of past convictions in that field as well.

As a legal matter, the U.S. Court of Appeals for the Third Circuit has suggested that such cases should be revisited. In Lee v. Glunt, a 2012 decision involving a 1990 arson case, the court held that a criminal defendant is at least entitled to an evidentiary hearing and additional discovery in instances where the state of the science has changed, and may also have a due process claim to a new trial.

It's important to recognize that these cases are not the result of prosecutorial misconduct, nor did expert witnesses - or anyone else - act in bad faith. Police and prosecutors investigated a possible crime and concluded, based in part on the collective scientific wisdom of the day, that a crime had been committed.

But a good-faith mistake is no less a mistake. We have an obligation to resolve errors to ensure the accuracy and fairness of the criminal justice system.

The public's faith in the criminal justice system can be enhanced if prosecutors get out in front of this issue by announcing clear criteria for reviewing past cases that may have been decided on the basis of outdated science. This has become routine in cases with DNA evidence, for example, and can be done through a variety of mechanisms. Many district attorneys' offices have established "conviction integrity units," internal teams that review convictions that, for one reason or another, are now under suspicion. The Innocence Project has provided external support in similar cases for years.

Another example is in Texas, which has established a state forensics commission that partners with prosecutors and defense attorneys to identify areas of concern, providing an unbiased and open process for creating a catalog of cases for review. The commission requires additional manpower and an additional budget, but it also provides a layer of protection for prosecutors, potentially innocent individuals, and our community as a whole.

As we rely more and more on scientific testimony, we must acknowledge that, for all their rigor, scientific explanations evolve, and we - as lawyers, judges, and juries - are not always as competent as we think in giving proper weight to scientific evidence.

The Dougherty case is a good place to begin. If expert testimony was discredited, the district attorney should determine through an impartial review that he is still confident of guilt beyond a reasonable doubt.

Such a review would send a message to the public that a mechanism is in place for justice to be done and that the judicial system in Philadelphia is administered in a way that the entire community can and should be proud of.

John Hollway is executive director of the Quattrone Center for the Fair Administration of Justice at the University of Pennsylvania Law School. jhollway@law.upenn.edu