NavList:

Jim Rives, you wrote:
"And, the graph shows some outliers, which I have left in the analysis.  Can't cherry pick.  
And, be consistent in the whole analysis approach.  Don't rush.  Record everything and keep it, even if you change them or don't use the data.  It might be just what you need looking back a year later. "

Good call. Yes, I agree. I mentioned a particularly difficult case in my Lunars workshop on Sunday. If you're shooting a quantity that you know should be changing in one direction, for example a lunar distance that should be slowly rising, or an altitude of a star that should be falling, it's tempting to toss an observation that doesn't match your expected trend. The known trend creates certain expectations about the way in which the observations should change that are hard to ignore.

Suppose you plan to measure the altitude of Polaris near the bottom of its little circuit around the north celestial pole as it has begun to rise on the right side. You shoot once and get 40° 48.6'. Then you shoot again after some minutes and get 40° 49.3'. After a few minutes more you shoot Polaris again and this time find 40° 48.9'. Wait -- it can't be lower... I've had this thought many times! It's tempting here to ignore the most recent observation and try again without even recording it. But the problem is that we don't know if this observation is actually worse. It may not be the "trend-breaker". It's possible that the previous sight was a bit too high. So keep 'em all and remain "agnostic" about the quality of each individual observation. :)

Has anyone here tried applying  Chauvenet's Criterion or Peirce's Criterion which I mentioned a few days ago? Any other tricks for tossing outliers?

Frank Reed