|
|
|
|
|||||
|
||||||
From: george huxtable (no email)
Date: Wed Sep 14 2005 - 19:51:46 EDT
When I wrote-
>"I should add that, just like Bill, I have no experience with the Davis
>model that Asbjorn was using, but Bill's comments tarred all plastic
>sextants with the same brush; and unfairly so, in my view."
Bill responded-
>It was not my intent to tar all plastic sextants, but merely pass along the
>feelings/observations of those more qualified then myself and with
>first-hand experience--like Celestaire and David Burch. Burch wrote in a
>PDF on his site http://www.starpath.com: "I would also propose ‹ as a broad
>generalization ‹ that using these procedures one should be able to obtain
>accuracy¹s of some 5 or 6 miles as a general rule with plastic sextants.
>Maybe better in some cases, maybe a bit worse in others." Celestaire's
>seminar handout claims 3 to 10 miles.
I wouldn't dispute an assessment of 5 or 6 miles for possible overall error
of an altitude taken with a plastic sextant on a small craft, in
less-than-perfect weather. But not all that error is due to the
plastic-ness of the sextant. Even with the most perfect and expensive
sextant, observations from a small craft in those less-than perfect
conditions will show errors of "a few minutes". These are unrelated to the
accuracy of the instrument itself, but dependent on observing conditions.
However, we were discussing a 10-minute error, in an on-land measurement
from a stable platform in Norway.
> In your own words while speaking of
>your Ebbco and IE checks, "...and it never shows a discernable change
>(reading to the nearest minute, which is all the instrument is good to)."
>Given that is only a minute, not five or ten, it is still greater by an
>order of magnitude than one would expect from a quality metal sextant.
That may indeed be true: I wouldn't dispute it. But it's not what we were
discussing, which was Bill's imputing of a 10-minute error in sextant
reading to variable index error in plastic sextants, in these words-
"If the error is relatively consistent within a batch, perhaps index error is
the culprit. I have no first-hand experience with plastic sextants, but
from what I have read IE appears to vary to much more within a batch, or
from batch to batch, than does a metal sextant."
I pointed out that IE, when properly allowed for, is NOT a source of error.
In my view Bill was simply relaying prejudice without first-hand knowledge
or experience of plastic sextants. It may be true, indeed that some models
are error-prone. Some may show errors in their scale divisions. I just
don't know (and neither does Bill). But I can state (and did) from
first-hand knowledge of two different plastic Ebbco instruments, that it
would greatly surprise me if defects in that make would give rise to a
ten-minute error in angle.
> Your
>results with the Ebbco are a testament both to the Ebbco and your abilities.
Don't get me wrong. I don't claim to be able to take altitudes to within a
minute; certainly not from my own small boat at sea, not with ANY sextant.
When I wrote about the Ebbco "reading to the nearest minute, which is all
the instrument is good to" I was referring to the precision to which the
scale could be meaningfully read, not the precision of the resulting sight,
which can be affected by many other factors.
>But if you are in the mood to shoot the messenger... <G>
Such messages carry more weight when based on first-hand experience, rather
than being the relayed opinions of others. So I hope that the messenger
will consider himself suitably shot...
>George also wrote:
>
>"Asbjorn described his technique, which as far as I could tell required
>adjustment of the index error to zero (using the grub-screws?) on each
>occasion, which appeared to me misguided. A better procedure would be to
>leave those grub-screws alone, after an initial setting, and just accept
>and allow for any subsequent index error: However large that index
>zero-error happens to be, if it's allowed for then it doesn't degrade the
>precision of the measurement AT ALL."
>
>I find this interesting. In my limited experience, adjusting for IE also
>affects side error.
That depends to a large extent on the layout of the adjustment-points of
the mirror. If they are delta layout, then it will happen as Bill
describes. If, better, in a "L" layout, then it won't.
>Again risking reprimand, what I take from the list is
>that a little side error is not a big problem, and may be useful.
>(Star-to-star or star-to-planet angular measurements?) I am not clear
>Asbjorn checked for side error after his index error adjustments.
>
>I suspect side error introduced by adjusting for IE would not account a 5'
>error.
I agree. And we are discussing a 10' error.
>The problem being I have no idea how to quantify side or
>index-mirror perpendicularity errors. My texts explain how to adjust those
>errors out, and check for side error with a star or horizon, but little
>more.
>
>The texts I have read first have one adjust for index-mirror
>perpendicularity. Then there are back-and-forth adjustments for
>horizon-mirror side error and index error until the index error is close and
>the side error is eliminated. One check for side error involves rocking the
>sextant to see if the horizon remains aligned. Being landlocked I use a
>faint star.
Fair enough. Or in daytime you could use a distant rooftop or a flat
skyline or a distant electrical power-line, with the sextant tilted.
The only reason I can think of for bothering to get the index error near to
zero is to simplify the arithmetic when subtracting it off. An index error
has no effect at all on the precision of the final answer.
But going back to Asbjorn's adjustment technique, what worried me was
allowing for backlash. If you make an index check, then an altitude
observation, ending each by turning the drum in the same direction, that
corrects for backlash automatically. If you indulge in the unnecessary
practice of setting the index error to zero with the grub screws, then you
would have to be similarly careful about the way the index arm had been
brought into that zero position. Asbjorn has recently cleared up the matter
by stating that he checked his index zero with his altitude observations,
but (I think) didn't state specifically whether he took precautions against
backlash.
>The questions arise:
>
>1. How (other than the arc check for rear silvered and block check for
>front-silvered mirrors) to check for index-mirror perpendicularity error in
>the field?
The distinction is NOT whether the index mirror is rear silvered or front
silvered, but whether or not the pivot line passes through the effective
reflecting surface, wherever that may be. If it does, then you can check by
looking at the reflected arc. What's the difficulty about doing that, in
the field? And having made the adjustment initially, is it an important
matter from then on? Unless the sextant gets dropped, that is...
>2. What errors will be introduced, and to what magnitude, if the index
>mirror is not perpendicular?
>
>3. What errors will be introduced, and to what magnitude, if the horizon
>mirror is not perpendicular?
>
>It strikes me that by using a star to adjust out side error, it may appear
>perfect, but still have problems in use if the index mirror is not exactly
>perpendicularity.
>
>Any thoughts or expertise you or the list could offer would be appreciated.
I would guess that any such misadjustments would have to be unfeasibly bad
to give rise to an error in measured angle of ten arc-minutes.
>Bill
George.
===============================================================
Contact George at ,or by phone +44 1865 820222,
or from within UK 01865 820222.
Or by post- George Huxtable, 1 Sandy Lane, Southmoor, Abingdon, Oxon OX13
5HX, UK.
|