Learning from the time and length redefinitions, and the metre demotion.

After discussing several issues in a future redefinition of the kilogram, this paper considers the lessons that one might have learned from the analogous redefinitions of the metre and the second. The progress of length metrology was slow and steady, from seven digits reproducibility with the 1889 X-shaped metre prototype, to nine digits with Kr lamps, to 11 digits with the 1983 redefinition of the metre using the speed of light. With laser cooling, the Cs clock improved to 15, now 16, digits (and so also astronomical distance measurements could improve). Laser-cooled ions, and now atoms captured and cooled in an optical lattice, enable accuracy capability of three different optical frequency references to exceed 17 digits, i.e. better than time itself. The optical comb and related techniques vastly simplify frequency comparisons. Such progress stimulates a new satellite experiment, the STAR Mission (Space-Time Asymmetry Research). The goal is to test at the 1E-18 level frequency shifts owing to spatial anisotropy, position, gravitational potential and boost. The onboard optical clock will use stabilization to a molecular transition in I(2) or HCCH or CO(2). The length etalons will be multiply redundant, with stability at the thermo-mechanical mirror motion limit. For a ULE glass etalon spacer (1987), I measure length creep approximately -1.5E-12/d, i.e. below 1E-14 over the 500 s satellite spin period.