Page 4 of Murdock's 1967 book Ethnographic Atlas: "The fundamental criterion for the establishment of clusters is the assumption that at least 1,000 years of separation and divergent evolution are necessary before two societies derived from a common ancestor are likely to develop sufficient differences to be treated as independent cases for comparative purposes."
He then gives a second principle: "No world sample should include any two societies located geographically so close to one another that diffusion is likely to have jeopardized the essential independence of their culture."
There follow other principles of sample selection from which his 412 clusters were constructed to serve as an initial sampling frame. This frame was later condensed into a sampling frame of 186 societies for the Standard Cross-Cultural Sample.
Murdock and White (1969) showed empirically that no world sample of cultures of useful size for statistically purposes is free of autocorrelation.