Over the past few years many proofs of the existence of calibration have be
en discovered. Each of the following provides a different algorithm and pro
of of convergence: D. Foster and R. Vohra (1991, Technical Report, Universi
ty of Chicago), (1998, Biometrika 85, 379-390), S. Hart (1995, personal com
munication), D. Fudenberg and D. Levine (1999, Games Econ. Behavior 29, 104
-130), and S. Hart and A. Mas-Colell (1997, Technical Report, Hebrew Univer
sity). Does the literature really need one more? Probably not. But the algo
rithm proposed here has two virtues. First, it only randomizes between two
forecasts that are very close to each other (either p or p + epsilon). In o
ther words, the randomization only hides the last digit of the forecast. Se
cond, it follows directly from Blackwell's approachability theorem, which s
hortens the proof substantially. Journal of Economic Literature Classificat
ion Numbers: C70, C73, C53. (C) 1999 Academic Press.