NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: The Online Nautical Almanac- beware!
From: Frank Reed CT
Date: 2004 Jul 4, 23:59 EDT
From: Frank Reed CT
Date: 2004 Jul 4, 23:59 EDT
George H wrote:
"Of course, 1800 is one of those years which divide by 4 yet which isn't a
leap year. The Online Almanac program gets that non-leap-year right for
1800, in not printing output for 29 Feb, and getting the day-sequence
correct as Thursday 27th, Friday 28th, Saturday 1st March. But somehow the
computation of Julian Day Number must have slipped here by a day, for dates
preceding that leap year."
Interesting. It's a rare thing indeed to see this sort of problem today but I suppose since software development is largely focused on future dates it's a little less surprising. I wonder if he's using some built-in library to calculate serial dates. I'm always suspicious of those new-fangled things...
And wrote:
"This is a bit ironic. I think it was Frank Reed who pointed out that when
Hamilton Moore produced his "Practical Navigator", a few years prior to
1800, but with predictions spanning that period, he failed to allow for
1800 being a non-leap year. That allowed Bowditch, in his own first
edition, to scoff at all the errors that ensued in Moore's predictions, for
dates after 28th Feb 1800."
Very ironic. That wasn't me, by the way, but while we're here I would like to thank whoever DID bring it up since it's a fascinating point. I have found it a great introduction to Bowditch for students. I tell them "he found the 19th century's version of the Y2K problem" and then take off from there.
And:
"The web page referred to above states clearly that "valid dates are in the
range 1950 to 2050", so those using the almanac outside that interval do so
at their own risk. Yet I have found the predictions to be usefully accurate
right back to the early 1800s."
A lot of the almanac data is certainly accurate over a wide range of dates, but there are problems nonetheless. The positions of the stars are inaccurate by a minute of arc or so by 1850 and worse farther back. I suspect that site uses a short time period model for precession which may be fine for the recommended timeperiod, but grows progressively worse over longer time periods.
And:
"I know that Ken Muldrew has been investigating David Thompson and other
travellers in Canada around this period, and one of his postings
specifically refers to data from the Online Nautical Almanac for October
1800."
Yes, and because of the (rather serious) problem with star positions, Reis's almanac should not be used at all for star positions outside its recommended date range.
And:
"I suspect that he never envisaged his predictions being
used for the purposes we are putting them to, so could quite legitimately
respond by pointing out that we are working outside his stated validity
period of 1950 to 2050. But I hope he will be more constructive than that."
Maybe he can clean things up if he has time. In the meantime, I've assembled my own online Nautical Almanac equivalent which I will describe in another post. It's valid over the time period 1750 to 2050. And unlike Reis's, if you try to get data outside that time range, it will simply fail -- and ungracefully, too!
Frank R
[ ] Mystic, Connecticut
[X] Chicago, Illinois
"Of course, 1800 is one of those years which divide by 4 yet which isn't a
leap year. The Online Almanac program gets that non-leap-year right for
1800, in not printing output for 29 Feb, and getting the day-sequence
correct as Thursday 27th, Friday 28th, Saturday 1st March. But somehow the
computation of Julian Day Number must have slipped here by a day, for dates
preceding that leap year."
Interesting. It's a rare thing indeed to see this sort of problem today but I suppose since software development is largely focused on future dates it's a little less surprising. I wonder if he's using some built-in library to calculate serial dates. I'm always suspicious of those new-fangled things...
And wrote:
"This is a bit ironic. I think it was Frank Reed who pointed out that when
Hamilton Moore produced his "Practical Navigator", a few years prior to
1800, but with predictions spanning that period, he failed to allow for
1800 being a non-leap year. That allowed Bowditch, in his own first
edition, to scoff at all the errors that ensued in Moore's predictions, for
dates after 28th Feb 1800."
Very ironic. That wasn't me, by the way, but while we're here I would like to thank whoever DID bring it up since it's a fascinating point. I have found it a great introduction to Bowditch for students. I tell them "he found the 19th century's version of the Y2K problem" and then take off from there.
And:
"The web page referred to above states clearly that "valid dates are in the
range 1950 to 2050", so those using the almanac outside that interval do so
at their own risk. Yet I have found the predictions to be usefully accurate
right back to the early 1800s."
A lot of the almanac data is certainly accurate over a wide range of dates, but there are problems nonetheless. The positions of the stars are inaccurate by a minute of arc or so by 1850 and worse farther back. I suspect that site uses a short time period model for precession which may be fine for the recommended timeperiod, but grows progressively worse over longer time periods.
And:
"I know that Ken Muldrew has been investigating David Thompson and other
travellers in Canada around this period, and one of his postings
specifically refers to data from the Online Nautical Almanac for October
1800."
Yes, and because of the (rather serious) problem with star positions, Reis's almanac should not be used at all for star positions outside its recommended date range.
And:
"I suspect that he never envisaged his predictions being
used for the purposes we are putting them to, so could quite legitimately
respond by pointing out that we are working outside his stated validity
period of 1950 to 2050. But I hope he will be more constructive than that."
Maybe he can clean things up if he has time. In the meantime, I've assembled my own online Nautical Almanac equivalent which I will describe in another post. It's valid over the time period 1750 to 2050. And unlike Reis's, if you try to get data outside that time range, it will simply fail -- and ungracefully, too!
Frank R
[ ] Mystic, Connecticut
[X] Chicago, Illinois
