The Canadian National Research Councilattaches a "relative uncertainty" of 2.5×10−11(limited by day-to-day and device-to-device reproducibility) to their atomic clock based upon the 127I2 molecule, and is advocating use of an 88Sr ion trap instead (relative uncertainty due to linewidth of 2.2×10−15).[26]See magneto-optical trap and "Trapped ion optical frequency standards". National Physical Laboratory. Such uncertainties rival that of the NIST-F1 caesium atomic clock in the microwave region, estimated as a few parts in 1016 averaged over a day