Thank you all for pointing out my third grade math error. I'm really not THAT stupid! Guess I was just rushing. Anyway, If I had done my math right my answer should have been that there would be a 2 minute error for a .5 degree dial plate rotation If one sets his sundial towards Polaris when it is due east or due west of the meridian.
The article in vol. 3 number 4 of The Compendium deals with this problem. On page 9 on the graph it looks like the resulting time error for a one degree turn is about 5.5 minutes and indeed varies in a bellcurve throughout the day. Would it be correct to infer that a .5 degree turn would result in a 5.5/2=2.75 minute maximum error (at 12 noon)? Also, would the amount of error change at different times of the year and at different latitudes? I guess all of this is acedemic because unless you are a good astronomer how would you know when polaris is on the meridian and when it is safe to use it to set a sundial. In my Sundial Owner's Manual, I tell my customers that the best and easiest way to set their sundials is to use the "time method". This method only works whith those dials that are perfectly designed, constructed, and leveled. The dial is set by rotating the level dial until it agrees with a reliable time source (I use a cordless phone and the number for time) and correcting for the Equation of Time and longitude. I state that this method is more accurate than using Polaris. Is this statement valid?
