SEARCH PARAMETERS FOR THE SMALL WORLD PROBLEM.

In "small world" studies successive men each send on a postcard packet to an acquaintance who may know a designated target person on a first-name basis. `In this analysis the likelihood of discarding the packet is separated from the chance the packet if sent reaches a man who knows the target. The former probability is assumed to be independent of the previous number of persons in the chain. The assumption is satisfied by the three extant sets of aggregate data. Then the latter probability is found to rise from zero to a modest plateau with the number of steps in the chain. Refinements and variants of the model are suggested. Funneling of chains through a few persons near the target is discussed. [ABSTRACT FROM AUTHOR]