NavList:

Hi all,

I must come down squarely on Jeremy’s side on this one, but do remind  that I have repeatedly preached the limitations imposed upon the ocean navigator by the condition of the sea horizon  – other than inadvertent errors in reading off the sextant or chronometer, which can to a degree be controlled by care, it is probably the most consistent cause of position error or non-conformity among a group of sights. Haze, glare, cloud shadow, low lying clouds, sand storms, abnormal temperature gradients, etc., all militate against the navigator, especially if unequally distributed as respect a round of observations. There are, in my personal experience, areas in this world, with specific reference to the Red Sea, where AM sun positions may place a vessel 16-miles East of the track and PM sights an equal distance to the West thereof – and in approaching Capetown at night, where I have sighted Green Point Light and Slangkop Light above the horizon at 60-miles by cross bearings, only to have them disappear completely within the hour. Science notwithstanding, refraction is a tricky devil against which eternal care must be exercised.

It has always been considered appropriate to observe stars, or even the sun at high altitudes permissive of back observation, on reciprocal bearings, i.e., N – S and E - W, or as otherwise convenient, so as to cancel out both horizon and sextant errors and, in the case of stars, to produce a box like fix, the size thereof being indicative of the uncorrected errors involved.  It was my custom, for what it’s worth, to observe a round of at least 6-stars, if available and conditions permissive, at 60-degree intervals to produce a fix – obviously this is not always possible, however, I do think it important to at least try to avoid taking all observations in one quadrant, or even semicircle.

Gary, in replying to Jeremy’s post has lamented, as I read it, the lack of statistical information necessary to evaluate the accuracy of celestial navigation before the advent of the sophisticated electronic means now at hand. He states .....

“During the history of celestial navigation IT was the most accurate navigation system available so there was no other data that the celnav positions could have been compared with to develop such an accuracy data set.”

It is not intended to question the correctness of Gary’s statement in the light of modern analytical and/or statistical methodology, of which I confessedly know little or nothing,

with particular emphasis on the nothing, but rather to invite attention to the means at hand to verify the accuracy of the celestial navigation practiced before the convenient availability of the new gear.

The ultimate test of the navigator’s skill, aside from keeping his vessel safely afloat, lies squarely in the quality of the intended landfall made – or, how far out was he when transferring from a celestial based position to one based on well charted terrestrial aids or, in more modern times, perhaps even Loran C, which was at one time excellent in the approaches to many US ports. Believe me, it was, and possibly still is, a matter of great satisfaction to make a perfect landfall “on time and on course” after some 20-days or so at sea. Did anyone make a statistical record of landfall accuracies? Some may have, but unfortunately, other than being filed in the good or bad memory department, I sincerely doubt it.

Distance actually steamed, as compared with true rumb line or great circle distance between ports of call, it seems to me, also can form a basis for assessing navigational accuracy, or perhaps more correctly efficacy. Some ships were kept very close to the intended track by frequent celestial fixes, thus minimizing steaming time and distance, while others were suffered to wander aimlessly to either side by sloppy navigational practices, excused by the allegation of an “unpredicted or unknown current”. Let’s take a couple of examples of what I mean ......

s/s African ___, New York to Capetown, Voyage 3:  True calculated Great Circle Distance = 6.764 nm; actual miles steamed by celestial navigation = 6,798 nm; difference = +34 nm.

s/s African ___, Capetown to New York. Voyage 4: True calculated Great Circle Distance = 6,764 nm; actual miles steamed by celestial navigation = 6,780 nm; difference = +16 nm.

s/s African ___, Capetown to Port of Spain, Voyage 3: True calculated distance by combination Great Circle + Rumb Line =  5,358 nm; actual miles steamed by celestial navigation = 5,380 nm; difference = +22 nm.

These are actual records of what I believe to have been a carefully navigated vessel solely by celestial navigation. Certainly they do not assess individual sight quality, but is it possible that such records might be used in retrospect to assess the overall quality of the navigation method employed? If so, they are potentially available in historical voyage abstracts, probably numbering in the many thousands.

There is no question but that some navigators were professionally enough interested in their skills to test their ability by observations for position in known locations. Were records made of such testing results, other than perhaps in personal diaries or workbooks? – certainly not, as there was no Internet, and no NavList, available as today for the instant and public exchange of information. I have posted on this List examples of the accuracy to be expected from various celestial navigation observations, as compared with known positions, always stating horizon conditions, sextant employed, and method of reduction; these postings have been casually accepted, generally without comment. Quite frankly, I do not believe there to be a great deal of difference in observing from your front porch, given that you have a view of the horizon, or the stable platform provided by a large ship in all but bad sea states; in fact, it may be easier from the ship, as a close inshore horizon seems more frequently hazy. With the instruments available to members of this List, it does not appear difficult that a project, leading to the establishment of a data base of observational error against know position, might well be undertaken.

As an aside matter of interest, during WWII ships in convoy were required to signal their noon positions to the Convoy Commodore by flag hoist. Given upwards of 100 ship convoys, any record of these positions, which I doubt exists, would provide a fertile field for the study of scatter in positional accuracy, if individually corrected for difference in convoy station. Unfortunately, during the war the keeping of diaries or navigational records was prohibited, although Admiral Doenitz, by way of the B-Dienst, knew the convoy positions only too well, so my navigational workbooks kept during the those years had to be destroyed. I do still, however, have 2 bound workbooks, totaling over 500 pages of closely packed, worked observations and related calculations done in the late 1940s and early 1950s, before going on active naval duty, for which I might be pleased to find a secure home. Copying is out of the question for me, as all the work was done with a 2-1/2 hard pencil and, although perfectly legible, too light for the conventional copy equipment to reproduce clearly.

Regards,

Henry

From: Anabasis75@aol.com <Anabasis75@aol.com>
Subject: [NavList 9479] typical standard deviation?
To: NavList@fer3.com
Date: Wednesday, August 12, 2009, 2:35 AM

George wrote:

 

"The deduced scatter, of one standard deviation about the fitted trendline, I
now make to be 0.53 arc-min in the case of the "Moon near LAM" set, and 0.38
arc-min in the case of the "Moon away from LAM" data set. That's no better,
and no worse, than one would expect from observations at sea from a large
vessel."

 

I wonder on what basis this statement is made.  I am not claiming to be any better or worse, than any other practiced navigator, but how can we know somewhere around 0.5 arc-min is what "...one would expect from observations at sea from a large vessel."  Is there some collection of data that would back this up?  I am just wondering how we can expect this kind of deviation from large ships as opposed to small vessels unless there has been some sort of study on ships of various sizes and under different observing conditions or a review of a variety of navigational logs.

 

Can any other navigator on this list give data that would support this kind of statement, even if the data isn't recent?

 

The reason I bring this up is that as I look through my recent navigational log, I notice that the most critical aspect of shooting a star is seemingly never mentioned (at least far less than sea state, large ships, and anomalies in dip).  This factor is the quality of the visible horizon!  The horizon varies with time, azimuth, and circumstance every time I shoot.  I have shot with a crisp horizon in one direction, and a fuzzy horizon in another quadrant at essentially the same time.  The horizon's quality at a given azimuth, affect my sights far more than any other shooting condition.  The quality of the horizon greatly changes the accuracy of my star fixes which is the only measure I truly care about.

 

Given the nearly endless variations of observing circumstances for any sight using the visible horizon rather than a bubble or other artificial horizon, I find it hard to justify stating in all but broad terms what magnitude of scatter we can expect from a given type of vessel.

 

Jeremy

---