NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: Advancing a position circle. was : [NAV-L] Position from crossing two circles
From: Alexandre Eremenko
Date: 2006 Jun 16, 03:10 -0400
From: Alexandre Eremenko
Date: 2006 Jun 16, 03:10 -0400
Dear George, You seem to discuss an interesting mathematical problem but I am not sure that I could extract the exact statement of the problem from your message. Apparently you ask this: Suppose we have a circle on the surface of the sphere. Then we move each point of this circle by a fixed distance in the SAME GIVEN DIRECTION. Will the resulting figure be a circle? 1. This problem is not well stated until you specify exactly what a "given direction" means. In navigation, we usually understand by "direction" the angle from the meridian passing through the point where we stand. Say, N direction or NW direction. But troubles happen near the poles. At a pole itself, this notion of direction has no sense at all. Near the pole, it has a sense, but this sense is different from what we usually mean by moving in a given direction. Suppose you are 1 mile from the N pole, longitude 0. And you move 10 yards West. Another guy starts at longitude 180, also 1 mile from the pole, and also moves 10 yards West. Do you think you and this guy move "in the same direction"? If this is what you really mean by "the same direction", the shifted circle will NOT be a circle anymore. And the size of the circle is irrelevant here. (Relevant is how far it is from a pole). Consider what will happen to a circle of radius 1 mile, centered 1 mile from the pole, if you move each point of the circle 1 mile to the West (that is 1 mile along the parallel passing through this point). The circle will become terribly distorted, the part near the pole will become a spiral winding infinitely many times around the pole. There are a few exceptional circles which will remain circles, but they are really exceptional. For example take a parallel. If you shift every point of it some distance N or S (or even NW) you will obtain another parallel, which is a circle. 2. Now we can ask the question: can one define rigorously, without using meridians, or other arbitrary lines substituting them, what does it mean that "every point of some circle is shifted in the SAME DIRECTION? The answer is NO. "Same direction" cannot be defined consistently for different points on the sphere. (This was essentially discovered by Gauss). Whatever system of lines you use to measure the direction, these lines will meet at some points and cause the troubles described above. 3. Now back to Navigation. If you are well away from the poles, a piece of the line of position near you is also away from the poles. If you shift every point of this piece of the position line by some moderate distance (=MUCH SMALLER that the distance from the poles), in the "same direction", as measured with respect to meridians, the result will be APPROXIMATELY a piece of a circle. Almost never will it be an exact arc of a circle. 4. Usually, this piece of a circle can be safely replaced by a piece of a straight line on a Mercator map. (As we almost always do in practice). 5. Now, when exactly is the use of a circle of position justified? When the radius of this circle is small. (That is you observe a body close to zenith). If you are far from the pole, you can safely shift each point of this small circle, say 10 miles NW and the resulting figure will be a circle very nearly. But not exactly. 6. The difference between "very nearly" and "exactly" makes sense only in pure mathematics. We know that the Earth surface is not exactly a sphere, and so on. The only practical question to ask about these idealizations is the magnitude of error we make using an idealization. If your circle of position has radius 60 miles, and you are at latitude 40, you can shift each point of your circle 10 miles W and assume that the result will be a circle "for all practical purposes". 7. Conclusion. To the mathematical problem stated in the beginning the answer is "no". The shifted line of position will not be a circle. In practice, if the original circle is far enough from the poles the shifted line of position will be a circle very nearly. Alex.