Arthur, I think in the last part of your message you might be asking when does the effect, if considered at all, become inconsequential? Perhaps the attached graph may provide some information, the equation used came from Meeus, Astronomical Algorithms (15.4). I hope the file size, 8k, doesn't prove to be a pain for anyone, I'm assuming it won't be but please feel free to flame and rave if it is.
I would guess that on whole the variation in atmospheric conditions would make it a rather difficult measurement to make and at the very least one involving a number of variables. Luke Arthur Carlson wrote: > Atmospheric refraction deflects the rays of the sun about 1/2 deg near > sunrise and sunset, which should result in about a 1-1/2 min error in > the reading of a sundial. Is this effect easily observed? At what > height above the horizon does this effect start to become noticeable? > > Art Carlson Attachment converted: MAC Hard Disk:REFRAC.GIF (GIFf/JVWR) (00005E5B)