NavList:

Dear Frank,

You wrote:

“….if you average four or five in a row, you can expect accuracy about four times better or +/- 0.1'.

Can lunars be better even than that? Probably not by much, but some rare observers have exquisite visual acuity and coupled with an excellent sextant and excellent observing conditions, it's possible to increase accuracy by another factor of two or even five. Mordris Fersters, in his NavList posts, has reported some high accuracy lunar statistics which strongly suggest he has that rare level of exquisite visual acuity.”

I think your “diagnosis” regarding me is slightly exaggerated. J

I have never claimed that it is possible to achieve higher accuracy than you had mentioned. When I wrote about 0,05’ margin, I was talking about deviation within one set of observations. Let’s suppose I have measured 6 constant angles (for example, Star to Star distance) and my sextant readings are:

30° 10.0’    

30° 10.1’

30° 10.0’

30° 10.1’

30° 10.1’

30° 10.0’

Standard deviation will be 0,05’ (it shows how much individual angle differs from the mean in this set).

Is this my final accuracy of the real angle? Definitely NO. It only reflects range of the errors between the measured individual angles.

Standard deviation in my lunar distance and star/star distance sets is:

- typicaly 0,05’…0,1’;

- sometimes 0,15’;

- rarely 0,2’;

- very, very rarely more than 0,2’.

These values have been obtained with 7x power scope.

After thousands of lunars I have taken, after finetuning my observational skills, I can say that the final error depends very largerly on the instrument (including, of course, the power and quality of the scope).

One thing that must be taken into consideration: when we take a single sight, we actually have to measure 2 angles (the second one is measurement of index error, no matter how we do it). Both these angles can accumulate some errors. Therefore, if the manufacturer has defined sextant accuracy 0,2’, it does not necessarely mean that under ideal conditions the error will be within these limits. In fact even 0,4’ error would be within the limits of manufacturer. Of course averaging of multiple sights will help, but not all instument errors are random. Of course there may be some “lucky” angles when all the errors are canceled out. But in general I would expect lunar errors for above mentioned instrument in range 0,2’…0,4’. I am collecting data and in future I will publish my SNO-T sextant results in more detailed way.  

In other post you mentioned zeroing out the index error. I know that you regulary recomend this method and I agree with this regarding traditional celestial navigation. For lunars (if we talk about expert level lunars) I would recomend check IE before and after every set of observations. I have noticed that even good metal sextants can slightly change IE. For example recently I took some lunar distances under very severe conditions. It was very cold: -27°C (-16°F). Index error had been changed some thenths of the minute of arc. For lunars it matters.

Modris Fersters

PS:

Many thanks to all NavList members (and to you, Frank, especially) for publishing so interesting posts. Every day I start and end by opening NavList to look for something new. I myself rarely post because my studies are connected with subjects that are not interesting for the most of the members. For example, I am now studying prealmanac era (before 1767) lunars. I have found in Cambridge arhives original calculations (those that took about 4 hours to calculate) made on board real ships. I am so excited!