"John Davis" <[EMAIL PROTECTED]> writes: > I have a question/challenge to all you sundial designers: what is the most > accurate design for a Standard Time dial? > ... > As a starter, the "Singleton" dial recently discussed here would seem to be > a reasonable candidate. It's main limitation, common to all dials which > incorporate an EoT correction, is that it is drawn for a some MEAN EoT > curve, and no allowance is made for the leap year cycle and the other minor > variations. Is there some geometry of dial plate and style which minimises > the time error caused by small year-to-year variations in the mean daily > declination? If this is achieved, then the small change in the EoT over a > single day may be allowed for.
The maximum rate of change of the EoT is about 30 sec/day toward the end of December. Averaging over leap years can be done to make the chart "wrong" by at most half a day, or 15 sec. The diameter of the sun is 0.5 degree, or 120 sec of time. Before you worry about the leap year problem, you first need to find a way to locate the center of the shadow edge 8 times more precisely than the degree to which it is smeared. We (e.g., John Carmichael and I) have discussed here some designs which might be capable of this accuracy, but they tend to be a bit hard to use. If you insist, one possibility is a camera obscura with a slit (ideally oriented parallel to the Earth's axis). This gives a sharp line image of the sun, which can be used to read the time from a series of date lines like we have been discussing. If you're really worried about leap years, you can pile four years' worth of dates on top of each other. The other approach advocated by some, namely determining the EoT directly from the declination, rather than the date, will always suffer near the equinoxes. For example, if you determine that the declination is 23 deg 11 arcmin +/- 15 arcmin, the EoT can vary over a range of 11 minutes! --Art Carlson
