Our Knowledge of G Gets Worse, Then Better

The precision with which we know a fundamental constant usually gets better with the passing years. Just the opposite, however, was happening recently to G, Newton's gravitational constant. In 1987, CODATA, the international arbiter of metrology, was quoting an uncertainty of $±128$ parts per million for G, the most poorly known of the fundamental constants. But then things got even worse. Several recent measurements have given values many standard deviations outside these limits, forcing CODATA to increase its assigned uncertainty to $±1500$ ppm for the latest compilation.