In March 2002, physicist Rusi Taleyarkhan announced in the journal Science that he had triggered nuclear fusion in a tabletop experiment using nothing but sound waves, a claim that upended decades of assumption that fusion required machines the size of small governments. His method relied on sonoluminescence, the odd process by which imploding bubbles emit tiny flashes of light and heat, dubbed a star in a jar. Horizon follows the fallout: Oak Ridge scientists Mike Saltmarsh and Dan Shapiro rerun the experiment and find no fusion, Taleyarkhan's team publishes a second paper two years later insisting the effect is real, and the argument stalls on one unresolved measurement, whether neutrons and light flashes occur at the same nanosecond. To settle it, Horizon commissions sonoluminescence specialist Seth Putterman to repeat the experiment with nanosecond-precision timing that had never been attempted before. His result is negative, with no neutron detected close enough to a light flash to count as fusion, though Taleyarkhan points to equipment differences as a possible explanation.