There's a famous example of how improvements in understanding of burn patterns resulted in concluding that a bunch of people were falsely convicted for arson:

Due to the extensive publicity the case received, and because the murder charge carried a potential death sentence, the prosecution hired Lentini and John DeHaan, who had coauthored a fire investigation textbook, to evaluate other theories of how the fire may have started. One possible explanation was that one of the children, playing with matches, had ignited a sofa.

Fortunately, two doors down from the Lewis’ residence was an almost identical house. Lentini and DeHaan received funds and permission to furnish that house with the same type of furniture and carpeting as in Lewis’. Then they wired the structure with sensors, lit the sofa on fire and recorded the results. Within minutes the house was an inferno, due to a flashover. A flashover occurs when a burning object generates hot combustible gasses that ignite and engulf an entire area in flames.

To the general amazement of everyone involved, Lentini and DeHaan discovered the same burn marks on the floor of the test house that prosecutors thought indicated arson. But rather than having resulted from a liquid accelerant, the marks were caused by flashover. Prosecutors quickly dropped the charges. “That case opened my eyes,” Lentini said. “There were all these rules of thumb you can find in the literature at the National Fire Academy that are just wrong.”