According to
Levin, Gilbert V., and Patricia Ann Straat. "The case for extant life on Mars and its possible detection by the Viking labeled release experiment." Astrobiology 16.10 (2016): 798-810.
Three naturally sterile soils (Moon, Surtsey, and one Antarctic sample) that tested negative (i.e., the active and control sequences were essentially the same) showed the validity of the LR [Labeled Release] in not giving false positives (Levin and Straat, 1976b).
The referenced 1976 paper agrees: during the design and testing phases of the labeled release experiments (so before it was done on Mars), the false positive rate of the LR experiment was tested on only three known-negative soils. (There is also no mention of these tests being blinded, as far as I can tell the experimenters were aware of which soils were known-negative.)
Using naive statistics, that gives an expected false positive rate of $1-\frac{3+1}{3+2}=\frac{1}{5}=20\%$.
Why did the designers of the LR experiment, knowing that they would have to defend any positive result against accusations of being a false positive, choose to study the false positive rate with so few known-negative soils?