Polaris , the nearest and brightest Cepheid , is a potential anchor point for the Leavitt period-luminosity relation .
However , its distance is a matter of contention , with recent advocacy for a parallax of \sim 10 mas , in contrast with the Hipparcos measurement of 7.54 \pm 0.11 mas .
We report an independent trigonometric parallax determination , using the Fine Guidance Sensors ( FGS ) on the Hubble Space Telescope .
Polaris itself is too bright for FGS , so we measured its 8th-magnitude companion Polaris B , relative to a network of background reference stars .
We converted the FGS relative parallax to absolute , using estimated distances to the reference stars from ground-based photometry and spectral classification .
Our result , 6.26 \pm 0.24 mas , is even smaller than found by Hipparcos .
We note other objects for which Hipparcos appears to have overestimated parallaxes , including the well-established case of the Pleiades .
We consider possible sources of systematic error in the FGS parallax , but find no evidence they are significant .
If our “ long ” distance is correct , the high luminosity of Polaris indicates that it is pulsating in the second overtone of its fundamental mode .
Our results raise several puzzles , including a long pulsation period for Polaris compared to second-overtone pulsators in the Magellanic Clouds , and a conflict between the isochrone age of Polaris B ( \sim 2.1 Gyr ) and the much younger age of Polaris A .
We discuss possibilities that B is not a physical companion of A , in spite of the strong evidence that it is , or that one of the stars is a merger remnant .
These issues may be resolved when Gaia provides parallaxes for both stars .