Only once before have we conducted a robotic search for extraterrestrial life. The Viking spacecraft carried three experiments to search for life in martian soil samples (71), implicitly adopting a metabolic definition. But instead of finding unambiguous evidence of martian biology, Viking appears to have encountered unanticipated nonbiological oxidant chemistry (71, 72). The Viking gas chromatograph mass spectrometer (GCMS) failed to find any organic molecules (released in stages up to 500°C) in the martian soil at the ppb to ppm level (73). The GCMS provided a de facto search for life that implicitly assumed a biochemical definition: no (detected) organics, no life. In effect, a metabolic search for life yielding ambiguously positive results (71) was undercut by the negative results of a search based on biochemistry.
With the benefit of 25 years' hindsight, we suggest a number of lessons to be learned from the Viking experience (ref. 7; in the search for life on Europa). (i) If payload limits permit, a remote search for life should employ experiments that assume contrasting definitions of life. (ii) If only one life-detection experiment can be flown, the biochemical definition likely trumps other definitions. (iii) It is crucial to establish the geological and chemical context within which biological experiments will be conducted. Had the presence of the martian oxidants already been demonstrated, different biology experiments would have been flown on Viking. (iv) Life-detection experiments should provide valuable information even if they fail to find life. (v) Nevertheless, exploration often cannot be hypothesis testing. Much of what we do in planetary missions is simply exploration.