Text from page 92 of SKY & Telescope May 1997 by Bradley Schaefer
...seen making it unclear what advantage the polarizer offers those lost at sea. Under a cloudy or even hazy sky (where no blue is apparent), the cordierite fails completely since the light has been so thoroughly scattered that all polarization is lost. With partially cloudy or twilight conditions the cordierite might work, yet the Sun's position is more easily apparent from the sky brightness distribution. So there are no conditions known under which a cordierite sunstone could provide any extra utility for navigation.
A POLARIZATION SUNDIAL
The sunstone proposal just outlined presumes that the Vikings observed the sky's polarization to find the direction north (if they knew the time of day). But this idea can be turned around, so that observations of the sky's polarization can yield the time of day if modern Vikings know which way is north. When the sky is viewed through a polarizing filter, a broad dark band stretches across the sky 90 degrees; from the Sun. When viewing toward the north, a perfectly horizontal dark band tells you that it is local noon. A perfectly vertical dark band ells you that it is either 6 a.m. or 6 p.m.--an ambiguity that generally is easy to resolve. The daily rotation of the dark band tracks the Sun's hour angle just like a shadow . In theory, we could merely view the north sky through a polarizing filter, rotate the filter until the dark band's orientation is apparent, and deduce the time from its angle to the horizon. Such a procedure will have poor accuracy, since there is no accurate means to determine the band's angle. An alternative idea would be to spin the filter until the sky is darkest, then measure its rotation angle. But such a procedure requires you to compare brightnesses separated by time, a skill most humans find very difficult. A much more exacting procedure would be to attach a so-called half-wave plate (which rotates the plane in which polarized light vibrates) in between the sky and the polarizing filter. This plate should cover only a portion of the sky visible through the filter. When the assembly is rotated, the unobstructed sky and the sky seen through the half-wave plate will brighten and darken out of phase. At some orientation the two regions will match precisely, and the human eye is very good at determining when adjacent fields have identical brightnesses. Under routine conditions the angle of rotation can easily be measured to better than a half degree. Detailed calculations for the threshold of detectable contrast show that half the time people with normal vision can match the fields to within 0.043 degrees; when using care. This angle corresponds to 10 seconds of sundial time, and indeed I find noticeable changes after 20 seconds in practice. You can design a polarization sundial to accurately tell time on any clear day, deep into twilight. First, make a sighting tube that points up at the north (or south) celestial pole. Then mount a half-wave plate over part of a polarization filter in a rotating turret at the end of the tube. Inscribe a scale where 6 a.m. to 6 p.m. covers 180 around the turret. After careful calibration, fasten a pointer to the tube for reading the local apparent solar time. The price paid for the accuracy such a device provides is that there will be two orientations (separated by 90 degrees;) that win have the brightnesses match. his ambiguity can be resolved by always rotating the device counterclockwise with the half-wave plate re-...
When you are looking toward the north through a polarizing filter (in this Case, through Schaefer's sundial), the sky appears to have a certain bright nest depending on the filters orientation. An intervening half-wave plate (narrow strip) rotates the light's polarization orientation. Rotating the plate until the brightnesses match allow Schaefer to read the current time from a properly positioned scale on the tube.
Left: This polarization sundial makes use of a half wave plate (a strip of cellophane) and a polarizing filter The half-wave plate is rotated until the apparent brightness of sunlight passing through it matches that of light passing through the filter alone. Above: At the back end of this sundial observers view the reflection of incoming sunlight off a shallow layer of water in the tube. The reflected light is polarized, eliminating the need for a polarizing filter.