The Labeled Release experiment looked for signs of microbial metabolism in soil samples in 1976. The general thinking was that the experiment had found geological not biological activity
However, the new study approached things differently. Researchers broke the data into sets of numbers and analyzed the results for complexity. What they found were close correlations between the Viking results' complexity and those of terrestrial biological data sets. Based on this they concluded that the Viking results were more biological in nature than just geological processes."
Link to Original Source