John Carmichael wrote: > You know Art, you really write GREAT answers!
That's probably sarcastic, but my ego refuses to entertain such a thought. My objective side would advise you to listen to all the advice from others who have actually built large dials. > I thought that I could use either or both of the methods > described by Mayall > pg.75-77. I'll call these" the concentric circle method" and" the apparent > noon method" Wouldn't both of these methods show true north very > precisely > if the vertical gnomon was very tall? They can. Even here you should consider sources of error. Near the equinoxes, for example, the change in declination during the course of one day could be enough to skew your north-south line by a good fraction of a degree, possibly more than the precision with which your dial can be read. > Once I know the true meridian, all of the right angles to it are easy to > mark using plane geometry. In geometry class I learned to construct a perpendicular to a line. The method can be rigorously proven to be exact, although the lines as they end up on my paper may not be. I recently learned how carpenters make a large rectangle: They lay out something that looks about right, then compare the length of the diagonals, then tweek things until they are really right. The point is that there may be methods which are not rigorous in the sense of high school geometry but are easier to use and may even give more satisfactory results. > Well, I will certainly take great pains to see that this does not happen! > Using water level as a guide, it is possible to produce a perfectly level > surface. )This is how the egyptians leveled the base of the > pyramids. If I > orient the azimuth of my gnomon using Mayall's methods, and I > know that the > face is level, then using simple geometry, it is easy to set it at the > proper angle or height. This will work, though it gets complicated if you need a slope for drainage, your site is deliberately on a hillside, you are installing a dial on a pre-existing surface, etc. I have given some more thought to the previous thread involving how to set a sundial using the time method to minimize errors, but anything like that is probably out of the question for a civil engineering sized dial. > But which day of the year is the best for doing this? Please > tell me if I'm > wrong in my thinking, but wouldn't you want a day where EOT=0 because then > you could quickly mark the timelines by a clock without corrections. (I > suppose you could also set the clock off by the EOT amount on the day of > hour line marking). This would give a longitudinally corrected dial. You > would also want to do it near the solstices when solar declination is > changing the least. Right? So the best day of the year would be Dec. 25. > What do you think Art? Is this correct? It doesn't seem any harder to me to take measurements at 3 min 46 sec past the hour rather than on the hour. Why are you worried about the solar declination? If it's a question of the change of the EOT during the course of your day of measurements, then December 25 is the worst possible day (15 sec in 12 hours). > >Isn't a giant protractor just a piece of non-stretchy string? > > No, a piece of non-strechy string is just drawing compass that hasn't been > nailed down yet! You can tell just how long ago that geometry class was! A curved scale is hard to make, and I don't see any real advantage over linear measurements. I would establish a few primary reference points, e.g., with the concentric circle method or EOT corrected shadow measurements, and then map several secondary reference points from these using a tape measure. Work down from there to the smallest scale you need. Always use measurements from at least three reference points to reduce errors (and also get an idea for how accurately your work is proceeding). I am presuming that you know enough analytic geometry to calculate the distances from the reference points and have a computer to help with the numbers. Keep us up to date on any large projects you start! --Art Carlson
