A question for the traditional ocean navigators?

[Skylark]

I’m fairly comfortable in reducing a sight and plotting a position line but my only experience has been as a back-up to GPS. Hence only ever practicing the skill, never used it in anger.
How inaccurate can the Chosen Position be for the process to still work? Say, for example, it had been overcast for several days, no possibility to take a sight. For what ever reason, your Dead Reckoning position is miles out. Does there come a point when the PL is effectively unsafe or do you simply use this to improve your DR position as the basis for the next sight and so on until a more reliable fix can be obtained?

 

[oldvarnish]

Not an expert, although I've done enough of it.

I would have thought it wouldn't matter what your chosen position was - you'd just get an enormous intercept which would be difficult to plot. But it should still work.

(I can't get my head around whether it would still work if your chosen position was in a different hemisphere, or the other side of the date line)

 

[dom]

It depends a bit on the system/accuracy of the sight reduction tables you are using. But as oldvarnish says you might need a massive chart/collection of charts if you choose not to start with a reasonably accurate AP (assumed/chosen position). For this reason the DR position is normally used, which is usually more than accurate enough for a sailing vessel, somewhat trickier for an aircraft.

Other than that you can use pretty much any AP as long as you are solving the spherical navigation triangle directly, which you will be if you're using a laptop or standard haversine derived tables. The only thing to watch out for if you choose a wildly inaccurate AP is the small error you will introduce by plotting a segment of a circle as a straight line.

 

[alant]

I’m fairly comfortable in reducing a sight and plotting a position line but my only experience has been as a back-up to GPS. Hence only ever practicing the skill, never used it in anger.
How inaccurate can the Chosen Position be for the process to still work? Say, for example, it had been overcast for several days, no possibility to take a sight. For what ever reason, your Dead Reckoning position is miles out. Does there come a point when the PL is effectively unsafe or do you simply use this to improve your DR position as the basis for the next sight and so on until a more reliable fix can be obtained?

Usual advice, is to use a CP within 30' of your DR.

 

[Cruiser2B]

I’m fairly comfortable in reducing a sight and plotting a position line but my only experience has been as a back-up to GPS. Hence only ever practicing the skill, never used it in anger.
How inaccurate can the Chosen Position be for the process to still work? Say, for example, it had been overcast for several days, no possibility to take a sight. For what ever reason, your Dead Reckoning position is miles out. Does there come a point when the PL is effectively unsafe or do you simply use this to improve your DR position as the basis for the next sight and so on until a more reliable fix can be obtained?

When doing the calc's longhand, the Assumed Position tends to be a fair distance from the DR, but when using programs or calculators we have a tendency to punch the DR in as AP and get correspondingly short intercepts. As a young subbie, assigned to redundantly plotting astro-derived LOPs over perfectly fine satnav-omega fixes, I learned to displace the AP well away from the DR. I usually used a nearish whole degree Lat/Lon for the AP, which made the chartwork (particularly on small-scale charts) a lot neater. Also found that the choice of AP didn't affect the end result - the fix always ended in the same place, within the normal limits of sextant and plotting accuracy.

Giving it a little more thought, we draw an LOP as a straight line, but it is actually an arc of a circle that could be thousands of miles in diameter. I suppose if you put the AP far enough away, the LOP will diverge from the arc enough for a measurable error in actual position. Don't imagine that would be a problem if you kept the AP within a degree of the DR.

 

[jdc]

I agree with all the above answers. I chose the CP such that Latitude is a whole number of degrees, and Longitude such that LHA is a whole number of degrees. This allows one to use pre-computed sight reduction tables which tabulate Altitiude and Azimuth for each degree of Lat and each degree of LHA, leaving one to interpolate only for one parameter, Declination, which is then easy to do. By this way the error is insignificant as the CP will be within 30' of the EP.

I wrote a slightly fuller explanation here: http://www.awelina.co.uk/sextant/reduction_tables/ReadMe.pdf, which describes the method and allows you to make your own pre-computed tables. I've used these in several voyages, without mishap so far!

 

[Birdseye]

Pure curiosity ( I've done my Ocean theory but never used a sextant in anger) how accurate are they in real life. When you compare your sextant position with the accurate one you get from a gps, typically how far out is it?

 

[jdc]

Pure curiosity ( I've done my Ocean theory but never used a sextant in anger) how accurate are they in real life. When you compare your sextant position with the accurate one you get from a gps, typically how far out is it?

I have found that the standard deviation (ie 1 sigma) of the position is 1.5 miles, with an apparently normal distribution.

I have found this both by observation, from which I have enough data points to say only that it's about 1.5 miles, and by convolving uniform (aka 'Top Hat') distributions from: sextant observation errors, sight reduction errors (atmospheric corrections mostly), ephemeris errors (principally due to truncation to 1' resolution), plotting errors (accuracy of drawing and plotting angles) and from the Walker trailing log (the accuracy of which I estimate to be about 1.5% of length run) which again gave me sigma = 1.5 miles, with a more or less normal 2-D distribution.

 

[AndrewB]

How inaccurate can the Chosen Position be for the process to still work? Say, for example, it had been overcast for several days, no possibility to take a sight.

I take it you are using the Mark St Hilaire method favoured by the RYA, which approximates a short section of an arc by a straight line. As Alant says, the traditional advice is ideally an Estimated Position within 30', i.e about 30 miles, of the true position, which works well within the scale of standard plotting paper. But even if your Estimated Position is a couple of hundred miles out in practice the results aren't all that bad even though the intercepts will be large, and may have to be plotted direct on your navigation chart rather than using plotting paper. If you are in serious doubt about your original EP, use the derived position as the basis of a new Estimated Position and repeat the calculations.

You might find it interesting to work through a text-book example with a really bad starting EP and compare it with the usual reasonably accurate EP.

 

[capnsensible]

I have found that the standard deviation (ie 1 sigma) of the position is 1.5 miles, with an apparently normal distribution.

I have found this both by observation, from which I have enough data points to say only that it's about 1.5 miles, and by convolving uniform (aka 'Top Hat') distributions from: sextant observation errors, sight reduction errors (atmospheric corrections mostly), ephemeris errors (principally due to truncation to 1' resolution), plotting errors (accuracy of drawing and plotting angles) and from the Walker trailing log (the accuracy of which I estimate to be about 1.5% of length run) which again gave me sigma = 1.5 miles, with a more or less normal 2-D distribution.

Wow. Have found Antigua a couple of times on the day I thought and in about the right 'should be over there' direction.

Using CP within 30' as alant pointed out.

Very Enjoyable!

 

[Skylark]

I take it you are using the Mark St Hilaire method favoured by the RYA, which approximates a short section of an arc by a straight line. As Alant says, the traditional advice is ideally an Estimated Position within 30', i.e about 30 miles, of the true position, which works well within the scale of standard plotting paper. But even if your Estimated Position is a couple of hundred miles out in practice the results aren't all that bad even though the intercepts will be large, and may have to be plotted direct on your navigation chart rather than using plotting paper. If you are in serious doubt about your original EP, use the derived position as the basis of a new Estimated Position and repeat the calculations.

You might find it interesting to work through a text-book example with a really bad starting EP and compare it with the usual reasonably accurate EP.

Thanks for all of the replies. You can probably guess that on cold Manchester morning I was day dreaming of warmer climes.

I always reduce sights manually using Mark S Hiaire / RYA method with AP3270 tables, I enjoy the process / maths.

That's a good idea, Andrew to look back through my nav log book (I'm an anorak) and use an intentionally bad Chosen Position.

30 mins seems fairly accurate? What if you've sat out gale hove-to for a couple of days? Would you be able to maintain a DR to that accuracy under those conditions?

 

[jdc]

...30 mins seems fairly accurate? What if you've sat out gale hove-to for a couple of days? Would you be able to maintain a DR to that accuracy under those conditions?

I've never been in that situation so don't know, but if hove-to I'd expect to drift at about 1.5 knots, which for 48 hours is only about 72 miles, and I suppose that one should be able to estimate drift to much better than about half the distance traveled. That should keep you nicely within 30 miles uncertainty.

Much more likely imho to give a substantial error (and what has led to my biggest errors in EP) has been when sailing fast, when one can easily cover 170 miles in 24 hours, so if you get no sight at all one day then the next opportunity for a sight could be more than 200 miles away, with associated greater circle of uncertainty. This is particularly so if using a trailing log as it skips a bit when going really fast.

But actually it's not often that you go that long between sights: I can think of only one or two days when at no time in a 24 hour period was there any possibility of getting a sight. What I have found is that if you pre-set the sextant for the expected altitude and are ready to scan the horizon in the expected direction of the sun whenever there is the slightest thinning of the cloud then a faint image of the sun can usually be seen through the clouds within 10 minutes or so of searching. A very quick adjustment to bring it down to the horizon, and you've a sight. A good telescope/eyepiece and filters are essential, but you don't need to wait for even the smallest patch of blue sky!

 

[capnsensible]

But actually it's not often that you go that long between sights: I can think of only one or two days when at no time in a 24 hour period was there any possibility of getting a sight. What I have found is that if you pre-set the sextant for the expected altitude and are ready to scan the horizon in the expected direction of the sun whenever there is the slightest thinning of the cloud then a faint image of the sun can usually be seen through the clouds within 10 minutes or so of searching. A very quick adjustment to bring it down to the horizon, and you've a sight. A good telescope/eyepiece and filters are essential, but you don't need to wait for even the smallest patch of blue sky!

+1

 

[Uricanejack]

I’m fairly comfortable in reducing a sight and plotting a position line but my only experience has been as a back-up to GPS. Hence only ever practicing the skill, never used it in anger.
How inaccurate can the Chosen Position be for the process to still work? Say, for example, it had been overcast for several days, no possibility to take a sight. For what ever reason, your Dead Reckoning position is miles out. Does there come a point when the PL is effectively unsafe or do you simply use this to improve your DR position as the basis for the next sight and so on until a more reliable fix can be obtained?

Others Have already confirmed the DR does not have to be accurate to get a good position.
If you have used your sextant to get a position using a satellite position as your DR.
You will get short intercepts. The position line is still only as accurate as the conditions allow.
In my opinion you have used your sextant for real or “in anger”.
I did it all the time when I had a satnav.

It’s always wise to check your position by observation.

Long ago prior to GPS and prior to there being a requirement for Sat Nav

This is what I used to do. 30 odd years ago

On one occasion going from Amsterdam to Hampton Roads at an economic speed.
After2 days of a storm. I came up with a position 5 miles nearer Amsterdam than we had been2 days earlier. We knew we hadn’t gone very far but that was depressing. The 2nd Mate agreed with me so that’s where we were.

We were several days late arriving and had been without a position for 3 or 4 days. The old man bugged me all morning. I told we were going the right way because I had been overtaken by an Aircraft Carrier but it was going to fast to Follow. We setup for a offset landfall aiming to pickup the coastline a bit North of The Cheasapeak and turn south. I suggested when we saw land we could put the boat down, go ashore to a pub and ask for directions . Capt didn’t think it was funny. Particularly since the 10 cm RADAR scanner blew of during the storm few days previously. And the 3cm display was NFG. We were using a 1O CM RADAR inter switched through a 3 CM Scanner.

I managed to get a few good shots of the sun over the course of the morning . Although noon was a bust, we got a couple of sights after noon to run back and came up with a reasonable position.

When the weathers bad stars are not going to happen, The Chief Officer would always tell me, His stars were much better than my noon’s,, I’d point out at least I got a position.

It can be frustrating and disconcerting to be days in bad weather without a position. Using DR. as the time goes by your DR gets worse.

There are a few things you can do. Which might help in a storm.

I liked to calculate my DR rather than plot my DR, I used traverse tables and with practice I was pretty quick.
Later I would use a calculator, A mercator sailing but you need mer parts for those.
I will have to find my Nories Tables some time.

Keep a regular log of your heading and if you can speed.
Ideally you want to steer one straight course and steady speed from noon to noon or noon to stars, Stars to stars and Stars to noon. Not likely in bad weather.

In bad weather on a sailboat your course will change over 24 hrs the closer you check the better your average. Do several DR’s just to keep up to date. every 4 to 6 hrs

You will be going Noon to Noon maybe 24, 48, 72 or more hours without an updated observation

There is a technique called a days work, I would use to calculate a DR over a long period. Its just a simple table of courses and speeds resolved into d lat D long and Departure then added up at the end.
Exactly like adding vectors.

In the morning I would be constantly watching the sky and horizon hoping for a brief glimpse of sun or horizon.
There was an old timer who used to post about how he could always get the sun. Even through cloud,
I could through thinner stratus or in breaks some times between fronts if they were far apart.
But forget it if it was raining you aren’t going to see through Nimbo Stratus.

Usually I would rush out with my sextant for a brief glimpse but have no horizon or I had a horizon and no sun.
Noon latitude would not happen.
Often it would be a hastily taken shot between clouds with a blurry vague horizon..
I would take as many as I could get over the morning. And use the ones I felt were the best if I got 3 or 4.
The 2nd mate would carry on the same through the afternoon and we would run them all up.

The official Noon position might not be declared until 1400 and was a bit of a fudge. But better than a DR.

The main thing was even on a good day, “never over sell your position” to the Captain or your self.
I would never claim an accuracy of 1 mile.
2 maybe under ideal conditions.
5 yeah probably
10 after a storm getting realistic more like 10 or20. But better than a DR.

When you are in the middle of the Atlantic you don’t need a position to 3 decimal places.
When you are passing the Sable or Bermuda you need a reality check on your accuracy. If its **** stay well away at least twice your worst guess on you accuracy.

Land fall is when you get nervous, or at least the Captain does. He would be bugging me all morning if we were getting near land.

Again you “don’t over sell your position” approach with caution and the possibility you might be wrong.
I liked to head for really big lights. In good visibility. When the visibility sucked an offset landfall.
RADAR helped if I could get within RADAR range I would soon figure it out.

I guess a hand held GPS with spare Batteries is good idea.
Or better yet, I don’t gout of sight of land anymore.

 

[dom]

Wow, what a wonderful stream of consciousness brought vividly alive with quirky detail and a big seasoning of common sense. Made me feel like I was sailing with you for a while, great post :encouragement:

 

[jdc]

The main thing was even on a good day, “never over sell your position” to the Captain or your self.
I would never claim an accuracy of 1 mile.
2 maybe under ideal conditions.
5 yeah probably
10 after a storm getting realistic more like 10 or20...

This is pretty consistent with a 1 sigma of 1.5 miles: You are not that likely to be within 1 sigma! Assuming 3 sigma as your circle of uncertainty is a good rule of thumb, which does equate to "5 yeah probably" quite well.

Now for a more controversial bit: I think 'noon' sights are a complete anachronism (although perhaps I'm on slightly shaky ground talking about anachronisms since the whole astro-nav thing is probably anachronistic anyway). Noon sights are a hang-over from the days when you couldn't trust the chronometer, so having at least a good latitude was essential. But with a good watch there's no advantage in a noon sight over any other, and plenty of disadvantage as it fixes the exact moment you have to take that sight - a moment when you might be busy, or a cloud has come over, or a rain squall has blotted out the horizon or any one of a dozen reasons. The reason it's still in text books is only because the authors of those books haven't engaged brain - ok there's some justification for including it in an addendum as an interesting historical quirk, but none at all in including it as the principal method.

 

[jwilson]

Pure curiosity ( I've done my Ocean theory but never used a sextant in anger) how accurate are they in real life. When you compare your sextant position with the accurate one you get from a gps, typically how far out is it?

If I had what I regarded as a good sight (you know when you do) or an average of several poor ones I used to usually get an LOP up to 2-3 miles out (occasionally 4) when using the cheapest Ebbco plastic sextant. This was when position confirmed by better means eg islands/lightships etc and the sights were just for practice. Later on a real passage the same cheap Ebbco put Rockall dead ahead pretty much exactly where I expected it to be, sights taken in a substantial seaway from a 34 foot yacht. This was before GPS existed....

By the time GPS did exist and I again started to make longer passages I found my eyes were no longer up to using the rubbish optics of the same old Ebbco and I bought an ex-WWII real sextant. With this I could pretty consistently get within 1-2 miles of a GPS position. Obviously occasionally you get a near dead-on LOP by accident.

 

[Uricanejack]

This is pretty consistent with a 1 sigma of 1.5 miles: You are not that likely to be within 1 sigma! Assuming 3 sigma as your circle of uncertainty is a good rule of thumb, which does equate to "5 yeah probably" quite well.

Now for a more controversial bit: I think 'noon' sights are a complete anachronism (although perhaps I'm on slightly shaky ground talking about anachronisms since the whole astro-nav thing is probably anachronistic anyway). Noon sights are a hang-over from the days when you couldn't trust the chronometer, so having at least a good latitude was essential. But with a good watch there's no advantage in a noon sight over any other, and plenty of disadvantage as it fixes the exact moment you have to take that sight - a moment when you might be busy, or a cloud has come over, or a rain squall has blotted out the horizon or any one of a dozen reasons. The reason it's still in text books is only because the authors of those books haven't engaged brain - ok there's some justification for including it in an addendum as an interesting historical quirk, but none at all in including it as the principal method.

Controversial? Not really. At best its just a running fix.
Observing the sun gets you a single position line except for a few very exceptional circumstances. The opportunity to cross with another body like the Moon. Or Venus does occur but tends to give a very similar position lines which don’t cross well.

Anachronism? To big a word for me. Were you ever a Chief Officer .You are correct there is no need to run to noon and do a Meridian Passage any two observations will give you a running fix, The closer the azimuth difference is to 90 the better.

The Noon Latitude has a few advantages the main one being its easy. KISS. There’s not much calculation involved and its convenient. The obvious disadvantage it doesn’t work when you can’t see the sun at noon.

Outdated, a matter of opinion.

 

[Malabar]

Thanks for all of the replies. You can probably guess that on cold Manchester morning I was day dreaming of warmer climes.

I always reduce sights manually using Mark S Hiaire / RYA method with AP3270 tables, I enjoy the process / maths.

That's a good idea, Andrew to look back through my nav log book (I'm an anorak) and use an intentionally bad Chosen Position.

30 mins seems fairly accurate? What if you've sat out gale hove-to for a couple of days? Would you be able to maintain a DR to that accuracy under those conditions?

Think about moving on to reducing the sights using basic trig formulae and a cheapo pocket calculator with trig functions (cost about £7). You only need a one EP for a group of sights. A bucketload of chosen positions, any of which might be 30nm away from your actual EP does nothing to improve the accuracy.

 

[Skylark]

Many thanks, Uricanejack, for a wonderfully helpful, informative and illustrative post. To all other posters, too. I was hoping to tease out the benefit of your collective wisdom.

I did the RYA theory course in about 2008 one to one with a retired master mariner. Because of my interest, he went well outside the syllabus. I did my qualifying passage in 2010 and managed to keep a fornoon, MP and afternoon running fix going for about 5 days. I enjoy the manual sight reduction process and wouldn't want to use a calculator. I last used my beautifully crafted sextant on a passage from Loch Ryan to Peel IoM, just to maintain some practice. I doubt that it will ever be my primary navigation tool but it's nice to know that I have the basics.

I always read tradtional navigation threads, air or sea, with great interest. Please keep giving generously of your unrepeatable experiences!