Abstract
Jon Elster reports that in 1940, and again in 1970, the U.S. draft lottery was challenged for falling short of the legally mandated ‘random selection’. On both occasions, the physical mixing of the lots appeared to be incomplete, since the birth dates were clustered in a way that would have been extremely unlikely if the lots were fully mixed. There appears to have been no suspicion on either occasion that the deficiency in the mixing was intended, known, or believed to favor or disfavor any identifiable group. If the selection was non-random in the way charged, Elster asks, was it unfair?