NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
From: Peter Fogg
Date: 2010 Dec 31, 14:03 +1100
Having
informed myself (when you're looking for a helping hand its always waiting
there, right at the end of your arm) that a "Gaussian distribution"
is quite simply a standard distribution, and that positive kurtosis appears to
be simply the opposite, and that leptokurtotic boils down to
excessive positive kurtosis (however that is defined) I may now make so bold as
to comment on the relevance of this jargon to the slope.
I guess we all know now what slope means; a methodology for the evaluation of a series of sights taken over a short period, but for the sake of anyone who may have tuned in recently, first a short diversion as to why the observation period must be short.
Its because drawing the calculated (=real) slope as a straight line is an approximation of a short portion of an arc (same as position lines). The extent of curve of that arc varies with the position of the body in the sky, but as a practical 'rule of thumb' the maximum period of time is assumed to be 5 minutes (for star/planet sights at dawn or dusk the available window of opportunity is, of course, another time-limiting factor).
Therefore the number of timed sights that can be recorded is limited by those 5 available minutes (mind you, if other factors allow you can go on taking sights then choose the 5-minute period you prefer - perhaps avoiding altogether any apparent outliers).
How many timed sights can you record in 5 minutes? If I have to record them myself its about 4 to 6, thus roughly one per minute - others may be faster. If I have a scribe then more, and I have posted here an example of 9 sights.
The relevance of this to whether the distribution is standard or not lies in the disproportionate effect 1 or 2 outliers can have. If both outliers lie to the same side of the slope, as they do in my 9-sight example, then they could lead to significant error if averaged blindly, and to adoption of a significantly erroneous slope if this is derived via linear regression.
In other words, because the population is so small a significantly non-Gaussian distribution or a leptokurtotic event, if this is the jargon you prefer, is always going to be likely.