NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: Lunars with SNO-T
From: Alexandre Eremenko
Date: 2004 Oct 26, 13:04 -0500
From: Alexandre Eremenko
Date: 2004 Oct 26, 13:04 -0500
George, Thank you for your interesting comments. I posted a second series of the Lunars made yesterday at Mon Oct 25 2004 - 22:46:20 EDT. Answering your implicit questions on the first series, the observations are listed in the order I took them. Sextant backlash of my sextant is negligible. It is guaranteed by the manufacturer to be at most 4", and I usually check it every time when I check my index correction, that is before each series of observations. The only possible explanation of that large error is some mistake during the observation (this was my first Lunar observation, after all:-) If you compare it with the second one, you will see that the second was much more consistent: my sighting technique improves. On my opinion, one should REJECT such an observation in such a series, without any hesitation. That looks just evident to me, without any scientific arguments. There is some difference between Lunar distances observations and doing science research. In my research, I would certainly pay great attention to any result that looks unexpected. But here we are not in the business of discovery:-) Also: my experience shows that I committ blunders, and I am always ready to accept this. But maybe some people don't:-) Alex. On Tue, 26 Oct 2004, George Huxtable wrote: > One has to be very careful. By excluding the odd point or two out of a set, > it's often possible to draw a straight line through those that remain which > gives a spurious notion of precision. Alex will have a better idea than the > rest of us about the precision he now expects from each of his own lunar > distances. > > What I'm getting round to saying, in a long-winded way, is that I rather > agree with Fred Hebard's point of view. Alex's point-4 may be classifiable > as a blunder, quite out-of-line with other observations, in which case it > could be plausibly rejected. Or it may be the result of statistical > scatter, in which case it should certainly be included. It's a borderline > situation, depending on one's judgment. > > In a previous incarnation, I was an experimental physicist, and this has > left me with a strong prejudice in favour of including "odd" observations, > rather than rejecting them. And indeed examining them more closely, in case > they have anything to tell us. Rutherford's discovery of the atomic > nucleus, for example, hinged on such non-statistical scattering of alpha > particles. Not that I expect Alex's lunars to win him a Nobel prize! > > If it was me, I would have included point-4. The resulting longitude, > averaging all six observations, would have given as answer a mean longitude > of 7' less than that of Alex's balcony: instead of 0.6' greater, as I make > it from averaging the remaining five. This would still be a remarkably good > result for a lunar, and something to be proud of. > > However, I respect Alex's decision to remove point-4, even though I would > have done differently. There are valid arguments either way. > > As a matter of interest, I ask Alex how he would have treated point-4 if > his observation happened to be 23.4' rather than 22.4'. It would still fail > his stated test, being out of a monotonic increasing sequence, so would he > then reject it automatically? I suggest that in that case it would be quite > wrong to do so. > > Please don't think I am claiming to be any sort of pundit on the > measurement of lunars. My own sextant is a cheap plastic one, useless for > lunars even if good enough for other observations from a small craft. Where > I do claim some experience is in the treatment of observational errors. > > George.