Friday, July 01, 2011

Neptune Completes First Orbit Since Discovery: 11th July 2011 (at 21:48 U.T.±15min)

Happy anniversary, Neptune!

According to those tables of numbers you get in books about the Solar System, the planet Neptune takes 164.79 years to travel once around the Sun. And Neptune was discovered 164.77 years ago as I write this post (1st July 2011).

This means our blue ice giant has still not made even one full journey around the Sun since being spotted and recognised for the first time by humans.

At some point this month, that 'first' orbit will be completed. The inhabitants of Neptune will be wryly noting the first anniversary of the inhabitants of Earth first realising we were looking at their planet.

Here on Earth, it's an excuse to celebrate a big round icy blue thing in space! Which is cool.

So I was wondering: when exactly is this anniversary? How do we know, and how accurately do we know? There are plenty of blogs and other sites claiming various dates, but few explain where their figures had come from, and none say how accurately they are known.

I thought this would be straightforward to find out, but it turns out it's not... and the quest was fascinating.

If you prefer raw facts to processes and explanations, here's the answer: the first orbit since discovery will be completed within about fifteen minutes of 21:48 U.T. on Monday 11th July 2011.

If you're wondering why this date is different to other dates you might find elsewhere, it's because I've thought it through in a lot more detail (I can get a bit carried away) and done it properly.

At least, that's what I think. Let me know if you disagree.

For those who'd like to read further, here is some suitable musical accompaniment:

To get a date for the completion of an orbit, a number of questions need to be asked:

When, exactly, was the planet "discovered"? (What does that even mean?)
Where was Neptune then? (What does that even mean?)
Where is Neptune now, and how accurately can we track it?
What is a "complete orbit"? Is the "first orbit" any different to any other orbit?

When was Neptune "discovered"?

Neptune is the first planet to be discovered that is definitely not visible with the naked eye. The first instruments capable of rendering Neptune visible were the telescopes made by Galileo in 1609.

Astonishingly – and this is 169 years before even the discovery of Uranus – Galileo himself observed Neptune on 28th December 1612, and recorded it as a star.

He observed it at least once more during the subsequent month. The motion of Neptune across the sky in one month is fairly small (about a sixth of a degree), but there has been speculation that Galileo was quite aware of it having apparently moved between sightings.

It's practically impossible that anyone could have seen Neptune before Galileo. But although he may have suspected something, he didn't consider it important enough to publish or investigate further, so it's fairly clear that he didn't believe it to be a planet.

[Edit 7/7: it is suggested that stars as faint as magnitude 8.0 may be visible under perfect conditions; perhaps even fainter for some individuals. Neptune can reach a peak magnitude of 7.78, which is 20% brighter than a magnitude 8.0 star; so it is theoretically possible that it could have been seen before the invention of telescopes. As there are many tens of thousands of stars with a very similar brightness, it would be unimaginably difficult to pick out even if you knew where to look. I'm not aware of anyone ever seriously claiming to have seen Neptune with the naked eye.]

Many further sightings were made of Neptune, always noting it as one faint star among many. The subsequent discovery story is full of intrigue and controversy. For me, what is fascinating is the new significance of mathematics: in fact, some have gone so far as to say that Neptune was discovered by mathematics before it was seen with a telescope.

Using Newton's Laws of motion and gravitation, astronomers were able to calculate the path Uranus should follow under the gravitational influence of the Sun and the other known planets. But Uranus gradually drifted away from this path. In 1845, John Adams and Urbain le Verrier both hypothesised that it was being pulled by something else. Independently of each other, they used the calculations to determine where this something else was, and suggested astronomers should look for a planet there.

In September 1846, Johann Galle, with his student Heinrich d'Arrest, made the first definitive sighting from the Berlin Observatory, locating Neptune less than a degree from where Le Verrier said it would be.

I disagree with those who say Neptune was "discovered" by mathematics. A hypothesis was made, and known laws of physics were employed to mathematically infer the most likely position of a new planet, assuming the hypothesis and the known laws of physics were valid for what was being observed. In this case, of course, they were absolutely valid, and the prediction led immediately (for Galle) to the discovery.

Johann Encke, who collaborated with Galle on later observations, said in 1846: "this is by far the most brilliant of all the planet-discoveries, since it is the result of pure theoretical researches, and in no respect due to accident."

Mathematics had established itself as a powerful tool for exploring the Universe.

Although Neptune had been seen many times before (including by people who were looking for it), and it had also been located using indirect observation and persuasive mathematical reasoning, it's clear that the first person to definitively observe it and know he was observing it was Johann Galle.

The date was the night of 23rd September 1846, and we can even pinpoint the time. It's in the Monthly Notices of the Royal Astronomical Society, Vol. 7, November 1846, page 155:

The planet's position was first recorded at a time of 12 hours, 0 minutes and 14.6 seconds, Berlin Mean Time. (It does seem strange to record the time to one tenth of a second! I imagine them taking all their readings diligently with a stopwatch.)

There were no national time zones in 1846, let alone any idea of a Coordinated Universal Time. Astronomers naturally used their own local mean time. The Berlin Observatory is located at 13º23'39"E, which means their local mean time would be UTC + 53 minutes 34.6 seconds.

(How accurately this was measured, or how accurately their stopwatch was synchronised to it, I do not know, but I would imagine the uncertainty could be reduced to a matter of seconds.)

So Neptune was first recorded at 23:06:40.0 U.T. on 23/09/1846.

Is that the "discovery" time? It must have been seen by Galle before that time... but was it recognised as being the planet they were looking for before or after the stopwatch reading was taken? Does it matter? Either way, it seems sensible to take this as the discovery time, to remember that the act of discovery is rather more fuzzy than the act of pushing of a stopwatch button.

But if what you're after is a time to let off blue fireworks or Neptune-themed party-poppers... we have a starting point to work with!

Where was Neptune at the moment of discovery?

The table reproduced above also tells you exactly where Neptune was recorded to be when it was discovered, but to be honest I don't know exactly what system they were using to measure it, so it's not really any use as it is.

What we need is system that tells us where planets are at any given time: an ephemeris.

The obvious place to go for this is NASA. Their HORIZONS ephemeris is as good as it gets. If anyone knows where the planets are, HORIZONS does. It's provided by the Solar Systems Dynamics group of JPL/CalTech (the Jet Propulsion Laboratory in the California Institute of Technology), and it's open to everyone.

Now what is best way to tell exactly where a planet is in its orbit, and how will we know when it has returned to that place?

I have to briefly be a little technical and long-winded while I make the case for employing ecliptic coordinates around the solar system barycentre. If you'd rather take my word for it (or if you know it all already), you can skip it.

First of all, there's no point in using the R.A. and Dec values that astronomers use to locate planets relative to the Earth, because we're interested in Neptune's orbit, and it doesn't go around the Earth.

Secondly, for a very similar – but more subtle – reason, we shouldn't use R.A. and Dec values from the Sun (known as heliocentric coordinates), because it doesn't really go around the Sun either. What Neptune orbits, to the extent that it can be said to orbit anything, is the Solar System barycentre, which means the centre of mass of the entire Solar System.

In many-body systems like the Solar System, there are no true periodic orbits. The planets, all the smaller bodies, and the Sun itself, all move in the gravitational field of each other. If a system is dominated by a single massive body like our Sun (which makes up nearly 99.9% of the mass of the Solar System), the other bodies tend to arrange themselves over time into approximately periodic orbits.

Normally, approximately is good enough, and we can just say Neptune goes around the Sun every 164 and-a-bit years. But if we want to know what date an orbit is completed, we're asking for (at least) an accuracy of 1 day in 164-and-a-bit years. Crude approximations are not the way forward.

Neptune's System:

Neptune is part of a small gravitationally-bound system of its own, comprising the planet itself, a large moon called Triton, and a dozen or more smaller satellites. Neptune is 5000 times more massive than Triton, and Triton is 200 times more massive than all the other satellites put together. For the most part, this system consists of Neptune and Triton both orbiting their common centre of mass, plus some flotsam.

The bound system of Neptune and its moons can be thought of as a single rotating thing, making its way around the Sun. And it is the centre of mass – the barycentre – that most closely follows a smooth periodic orbit.

As it happens, in the case of Neptune, this barycentre is only 74km from the centre of the planet. The planet travels at around 5.4km/s around the Sun and in very nearly the same plane, so wherever the barycentre goes, the planet will never be more than 14 seconds ahead or behind it. 14 seconds is a tiny amount of time compared to 164 and-a-bit years, so this is not something that will affect our results. Nevertheless, it's the barycentre that most closely follows a periodic orbit, so it's the barycentre we'll be following.

What Pulls Neptune Around:

The Neptune system (which I'll just call Neptune from now on) is around 4.5 billion km from the Sun. It is pulled around by a collection of massive objects within its orbit, all of which tend, on average, to pull Neptune continually towards the 'centre' of the Solar System. From Neptune's perspective, the giant planets – Jupiter at a mere 0.78 billion km and Saturn at 1.4 billion km – are all pretty close to the Sun. They may pull a bit to the left or a bit to the right, but for the most part, they pull in.

Even Uranus, at 2.9 billion km tends to pull inward, although often at more of an angle than the others. (It may at times be closer to Neptune than Jupiter or Saturn, but is less massive, and always has less gravitational influence than either of the gas giants.)

So the 'centre' that Neptune is pulled in towards is not the core of the Sun, but the centre of mass of all those objects within Neptune's orbit plus Neptune itself.

There are vast numbers of objects outside of Neptune's orbit, but (a) they are pulling in all kinds of directions, and these pulls will tend, on average, to cancel each other out; (b) they are usually a very long way away; and (c) they are tiny – the total mass of all of them comes to barely a dozen times the mass of Neptune's moons.

There is also a strange collection of objects that actually inhabit Neptune's orbit, herded by Neptune's gravity into two little clusters, 4.5 billion km ahead of and behind Neptune. They're curious animals; but they're also very very tiny and very far away.

The Barycentre:

It makes sense to ignore all the other little things, and say that Neptune is, on average, attracted to the barycentre of the Solar System, and is in an approximately periodic orbit around the barycentre of the Solar System.

Below is a diagram showing the motion of the barycentre relative to the Sun over a period of 50 years. (Note that we might more properly think of the Sun as moving relative to the barycentre, but that would be harder to plot). The most prominent effect is the 12-year cycle as the Sun does its tango with Jupiter. But it is obviously being pulled around in other ways too. (source)

Because of this motion of the Sun, tracking an orbit of Neptune relative to the Sun will give rise to some unnecessarily complicated relative motion.

Below are two plots of the distance to Neptune, the first measured from the Sun (in red), and the second measured from the barycentre of the Solar System (in blue), over six centuries from 1700 to 2300 (click on images to enlarge).

The periodic variation in distance associated with any elliptical orbit is clear in both plots. A closer look, however, reveals that the path around the barycentre is much more smooth. (Data from HORIZONS. Thanks to W. Folkner for suggesting this comparison.)

The Motion of the Solar System Through the Cosmos:

We've considered the motion of Neptune and its moons relative to their barycentre, and the motion of the centre of that system relative to the barycentre of the Solar System. What about the motion of the barycentre of the Solar System relative to the rest of the Universe?

I might say more on this in a future post, because I like that kind of thing. For now, I'll just say that that it's completely irrelevant to the motion within the Solar System. The Solar System is in virtually perfect freefall through its stellar neighbourhood and, like any object in freefall, unless there are appreciable tidal effects, what goes on within the system is entirely isolated from the gravitational effect of anything beyond it and unaffected by the nature of the path it follows. There aren't any appreciable tidal effects from outside the Solar System because the distances involved are far too large.

Some people like to think of the planets corkscrewing their way through space. If that's your thing, go right ahead, but it doesn't mean anything in physical terms. If there were an absolute frame of reference relative to which the Solar System could be considered to be moving, that might be useful in some objective way. But there are no absolute frames of reference in space. The choice of frame of reference is ours to make. For what we're interested in, the frame of rest of the barycentre of the Solar System is by far the best one we've got.

How to Measure Position Relative to the Barycentre:

Now we can come back to using the JPL ephemeris, HORIZONS, to establish a location for Neptune relative to the barycentre of the Solar System.

The settings I used are shown below:

These settings select a coordinate system centred on the centre of the Solar System, and use those coordinates to tell us where Neptune is. I've entered the time of the moment Neptune was first recorded (I've entered 23:06 and 23:07 on that date, as there's no scope for entering seconds; but we can always interpolate).

The full output can be seen here (or you can do it yourself). The figures that matter are:
X=25.74504003 A.U., Y=-15.4126128 A.U. and Z=-0.27570192 A.U.

These are the x-, y- and z-coordinates of Neptune. They tell us that if we want to get to Neptune from the centre of the Solar System (on the day it was found), we should go 25 and-a-bit times the Earth-Sun distance in the direction of the ascending node of instantaneous plane of the Earth's orbit and the Earth's mean equator at the reference epoch, then, with the North Star above us, turn right into the plane of the Earth's orbit and go 15 and-a-bit times the Earth-Sun distance that way, then turn down out of that plane and go a bit more than a quarter of an Earth-Sun distance, and there it is.

(The coordinate system is somewhat awkwardly defined, being based on the orbit of the Earth, but its axes are fixed relative to the distant stars. They are the axes of a frame of reference that is at rest with respect to the barycentre of the Solar System and would be inertial in the region of Solar System if the Solar System were not there. And that is all we need.)

So that's where Neptune was then.

When does it return to that point in its orbit?

All we need to do now is to find out when it returns to that point in its orbit, 164 and-a-bit years later.

Of course it will never return exactly to the same point... so we will have to settle for the next best thing, which is to find out when it returns to the same longitude. From the Earth, we can measure the celestial longitude of a planet, which is just how many degrees it has moved along the ecliptic, relative to the position of the Sun at the spring equinox. (The ecliptic is the path of the Sun across the sky.)

The HORIZONS coordinates will give us a celestial longitude for Neptune easily, because X and Y are based on the ecliptic. Using the figures quoted above, the longitude is the inverse tan of Y/X in the range 0º to 360º, which is 329º 5' 33.3"

From the centre of the Solar System, however, a longitude based on the ecliptic is not very useful. From the barycentre, the ecliptic is the current path of the Earth across the sky, but we're interested in the orbit of Neptune. The best general purpose longitude for the Solar System is the angle around the invariable plane.

Unlike the ecliptic or any other orbital plane in the Solar System, the invariable plane is absolutely constant. The paths of orbits of all the planets oscillate slowly about this plane, over tens or hundreds of thousands of years, but the law of conservation of angular momentum ensures that the invariable plane can never be changed by any of the complex dynamics of the Solar System. All orbits are ultimately paths around the invariable plane, with some additional movement above and below it.

The diagram below shows the relationship between the ecliptic (path of the Earth) and the invariable plane. The angle between them is exaggerated for clarity. I've marked the position of Neptune at the time of its discovery:

The longitude I want to use is the angle θ shown on the diagram. θ is the angular position of Neptune relative to the line of intersection of the planes.

The calculations are not worth reproducing in their entirety, but here's a little vector wizardry that looks pretty if you don't know what it means, but is enough to explain what I was doing if you do:

which gives θ = 41º 30' 35.86".

As you can see, this is so close to the angle of 41º 30' 35.6" on the diagram in red (on the ecliptic) that it's safe to say it isn't going to make much difference to the final outcome.

The next job is to use the ephemeris to locate Neptune at various times in July 2011, and find out when it returns to precisely this angle around the invariable plane. The result is:

11th July 2011, at 21:48 and 24.6 seconds U.T.

For the record, I also carried out this process using the ecliptic and using the current orbital plane of Neptune. The results are:

11th July 2011, at 21:50 and 27.7 seconds (ecliptic)
11th July 2011, at 21:47 and 31.7 seconds (orbital plane of Neptune)

These are so close, it clearly doesn't matter at all which definition of longitude you think is best. Agreement within a minute or two after 164.79 years is pretty good.

The output from HORIZONS for the relevant times is summarised below, so you can play around with it if you wish.

The reference to "Coordinate Time" in there tells us that it's the time when Neptune actually was at those coordinates, not the time when the light from Neptune reaches us. Light takes 0.1733 days (a little over 4 hours) to reach us from Neptune, as you can see from the output. If the distance has changed appreciably between two sightings, this can make a difference in terms of when we would actually see an orbit being completed. In our case, comparing the positions of Neptune in 1846 and in 2011 as measured from Earth, it's slightly more distant in 2011, but the difference is less than 30 seconds.

How close will it get to where it was when it was discovered?

At that moment, Neptune will pass within 1.5 arcseconds of its 1846 location relative to the barycentre. As shown in the image at the top of this post, this is less than the diameter of the planetary disk, so it will overlap its original place in the sky.

This corresponds to a distance of 32,460 km in the direction perpendicular to the invariable plane, or any of the other planes if you prefer. This is not the same as the change in the raw Z-coordinate from the ephemeris: the reason for this is that Neptune has moved in a little towards the Sun in 2011, and that alters the Z-coordinate for the same point in the sky. To be specific, it will be 347,750 km closer (0.0077% closer) to the Solar System barycentre than it was in 1846.

Combining these figures, we find that its closest approach to its discovery position is 349,260 km. These values are subject to uncertainty of the order of ±1000 km, as we'll see below.

How accurate are the data that I've used for this date and time?

It's probably fairly clear that giving these events to a fraction of a second is a bit silly. But it would be good to know how accurate we can expect them to be. For example, is it definitely on the Monday (11th)? Could it be out by a couple of hours?

Also, it's just good practice to establish a realistic degree of uncertainty.

The sources of uncertainty here are:

1. Our knowledge of the "time of discovery" of Neptune
2. A degree of arbitrariness regarding the choice of longitude
3. The degree of uncertainty in the HORIZONS ephemeris data we are using

The first one reflects the fuzzy interpretation of the word discovery, as discussed earlier. I don't know how to quantify the fuzziness, but given we have the exact time that its position was first noted, I'd be comfortable with ±10 minutes.

The second one has just been addressed, and we can see that it could introduce a vagueness (to add to our fuzziness) of a minute or two.

To find the uncertainty involved in the ephemeris data, I tracked down this report relating to the "JPL Planetary and Lunar Ephemerides DE405", which is the source of the data for the HORIZONS online ephemeris. The data was collected in its present form in 1997, and the report is dated 1998.

The report compares the positions of various bodies as given by DE405 with the positions from an older ephemeris, DE403. Assuming these are independent, the difference between them gives some indication of their level of accuracy. The newer DE405 was created with far more accurate information on the inner planets than its predecessor due to the various spacecraft that have taken precision equipment there, but the outer planets remain a little blurry.

Figure 8 at the very end of the report (reproduced below) shows that the differences between the two ephemerides for the longitudinal position of Neptune in the period 1846 to 2011 are around 0.1 arcseconds (or 0.1").

An arcsecond is 1/3600 of a degree. At the distance of Neptune, 0.1" corresponds to a little over 2000 km, and Neptune will cover that distance in around 400 seconds (nearly 7 minutes).

I think it's fair to assume that the majority of this difference will be due to inaccuracies in DE403 rather than DE405. But it's not unfeasible that a straight comparison could hide systematic errors common to both. Nevertheless I suggest that an uncertainty of ±5 minutes is reasonable.

Altogether, I believe we have around 15 minutes of uncertainty in the time of completion of the first orbit of Neptune.

With that, I'll give the final result (again):

The first orbit of Neptune since discovery will be completed within about fifteen minutes of 21:48 U.T on Monday 11th July 2011.

Other claims for this date:

12th July has been quoted widely for this event, including on Wikipedia which I cannot change as they do not permit "original research" (if this can be called that) or regard blogs as a notable source of information, and rightly so. Some have even gone so far as to specify a time on that day. This comes from heliocentric ecliptic longitudes as would be observed from the location of the Sun, and don't take into account the motion of the Sun during that period.

[Edit (7/7): The date of the 12th has also been quoted by a few informal NASA sources.]

10th July has also been quoted – the reasoning behind this is much more straightforward. The length of a year for Neptune is often quoted as 60190 days or 164.79 years (sometimes as 60190.03 days and 164.79132 Julian years). So some have simply use Excel or Wolfram Alpha to add 60190 days to the discovery date. If you do this, you'll get the 10th.

I've also seen the 8th July quoted, which is what you get if you use Wolfram Alpha to add 164.79132 years to the discovery date, without realising that this figure refers to Julian years (it is 164.79485 tropical years).

I wanted to see who was right and why, and it turns out that none of them were. (But then, who in their right mind would go to this amount of trouble anyway?)

Is the duration of this first orbit different from any other orbit?

I took a look at a few more orbits using HORIZONS ephemeris. I was pretty surprised at the amount of variation from period to period. The plot below shows how much the period changes over 26 orbits:

Some Neptunian years in this sample are over 50 days longer than others. There seems to be a periodic variation on a scale of several thousand years. And the period of this variation also appears to be increasing – the downward slope on this plot is steeper than the upward – which suggests the presence of more than one periodic driving force for the changes in year length.

As we know, Neptune is being pulled around by the other three giant planets in quite dramatic ways. As a result its orbit is subject to far more variation than that of the Earth.

The idea that "each year on Neptune lasts 164.79 times longer than a year on Earth" that we read in our favourite Solar System text books... it's not the whole truth, really, is it...

So now what?

Having done all that, the question now is: what shall we do on the 11th?

Any thoughts?

Here's a nicer link to this post:
Please quote this link if referring to results from this post.


PaulB said...

Thank you for this fascinating snippet.

What should we do. We should dust of our telescope's and get out there and observe the two distant ice giant's of the solar system.

Remember you our part of an exclusive club, if you manage to spot Uranus, and Neptune through small to medium sized binoculars.

Bob said...

Thanks Paul!

Spotting it through small-to-medium-sized binoculars is exactly what I've been attempting to do. I'm getting the sense that it might not be feasible (especially after reading forum posts) but I haven't given up yet!

It's passing very close by a little 5.4 mag star "e Aqr" according to Stellarium: 15' directly underneath. So now seems to be a good time. I'm going to take my small-to-medium-sized binoculars to somewhere with a clearer sky and try again...

AstroFloyd said...

I have a silly question - perhaps I'm misunderstanding something: if Neptune was discovered early on the 23rd Berlin time (12:00:15 am = 0:00:15 hours), shouldn't that correspond to late (23:06:40 hours) of the *22nd* UT?

AstroFloyd said...

Looking at the discovery story, Galle received Le Verrier's letter on the 23rd and they discovered Neptune the same night, so that must have been the night of September 23/24, which means that the date in MNRAS is wrong and you used the correct UT date/time.

Indeed, when I use the true heliocentric (rather than barycentric) longitude and your UT date/time of discovery, I find 2011/07/12, 17:30:43 UT as the anniversary moment, using VSOP.

Bob said...

Hi AstroFloyd,

Yes, it's definitely the night of 23/24. It's not uncommon for "Sep 23" to mean the period from noon on 23rd to noon on the 24th. Noon on the 23rd is Sep 23.0, and midnight is Sep 23.5. If you look through some of the other pages of the MNRAS (e.g. this one) you'll see this convention is in use. So there's no contradiction there.

Intrigued by your 17:30 on the 12th - could you explain how you got that? HORIZONS gives 21:03 using heliocentric ecliptic longitude; the correction to 'true' longitude would be of the order of one minute, so you're three and a half hours early...

Bob said...

On the subject of visibility, I am very happy to say that I saw Neptune this evening, using only a small pair of 10 X 25 binoculars.

A clear night, and a few miles from urban area (not brilliant location - some yellow glow), at 1.45am, and with Neptune at only 19º altitude.

It took a few minutes after finding the right star field for fainter stars to reveal themselves; Neptune itself was only visible with a bit of averting, but it was there, clear as anything.

I didn't know it was possible with such tiny binocs, but it can be done!

I'm especially pleased to have seen it during its first orbit since discovery :)

AstroFloyd said...

Hi Bob,

I used VSOP87 instead of DE405. The problem is that I don't have a routine for precession of heliocentric coordinates, so I used a simplified scheme. I'm not too surprised that my result is off by a few hours - without correction for precession is was off by ~1 year.

BTW, you can use HORIZONS to compute data for a given set of date/times by selecting "discrete time(s)" on the Time Span page. And you can actually specify seconds (down to ms I think) even though it's not documented, using hh:mm:ss.sss.

Bob said...

Ah, that's useful to know - thanks.

Bob said...

@Astrofloyd: even more intriguingly, Phil Plait gets a result consistent with yours for the heliocentric return: he uses a discovery time 1 hr and ~8min later, and finds the same longitude is reached at ~18:38, 1hr ~8min later than the time you gave. So it looks like you're vindicated! I'll have to redo my attempt again sometime.

Anonymous said...

Hi Bob and everyone else who has posted.
I love this topic. I respect your analysis using the SSB, but the consensus among my expert friends is that the heliocentric longitude is the more appropriate reference for a birthday. Nonetheless, I hope you had a good celebration today. In London, Ontario, the Royal Astronomical Society of Canada's local group celebrates with dinner at a local restaurant this evening. We might even sing an appropriate Nep-TUNE for the occasion.

Bob said...

Thanks for your comment.

Re heliocentrism: yes, that appears to be a common opinion. I don't understand why - I see it regularly thrown out as the orthodox view, rather than explained with any reasoning. I'd love to hear why you or anyone else chooses to take that view.

In terms of the observed nature of the orbit, it is extremely clear from the two plots (this one and this one) that the motion about the barycentre is a lovely smooth approximation to an elliptical orbit; and meanwhile, the Sun is doing its own wiggly dance in the middle.

In terms of the physical forces governing the trajectory of Neptune, it is also very clear that it is continually pulled towards something very close to the centre of mass of all the objects within its orbit, and can therefore be expected to follow something close to a Keplerian ellipse with the centre of mass of everything-within-its-orbit-plus-itself at one focus. As I explained above.

The only reason I can see for preferring a heliocentric longitude is if you were concerned with the moment when the Sun appears at the same point in the Neptunian sky relative to the background stars. I can see the attraction of that, but to me it seems trivial in comparison with the physical nature of the orbit itself.

Certainly if you're talking about completing a physical orbit, you'd think you'd be primarily interested in what it is physically orbiting.

So I respect these traditions and inspiring astrophysicists who stick to them, or who see it as a matter of taste or of definition (such as Carolyn Porco - here's what she had to say), I don't see the benefit in agreeing with them in the face of very good reasoning to the contrary.

Let me know if you or anyone else fancies presenting an argument the other way - I'd love to hear it!

jir1667 said...


I'm impressed with your detailed analysis. Don't let the likes of Carolyn Porco put you off, these sorts of calculations are certainly worth doing, if only for educational purposes. It may be a matter of definition but once that has been agreed, precise calculations can be done and a lot can be learned from that process. It's the way I like to work, have been for many years, and sometimes it does uncover limitations, or downright errors, in the work of the "professionals".

Anyway, the thing that caught my attention was the fact about how close Neptune returns to its original direction, a mere 1.5 arcseconds. I thought it would have been much greater. So I decided to check it myself and in the process see if I could reproduce your results for the time of Neptune's return.

I've used the ecliptic and mean equinox of J2000.0 as my fixed reference system and expressed the position (direction) of Neptune as an ecliptic longitude and latitude. There's no great advantage to using the invariant plane, unless it more closely matches Neptune's orbital plane. The ecliptic and equinox do move, but if we fix them at some epoch (J2000.0 for example) then there's no reason why we can't use them as the basis of an inertial reference frame.

Relative to the SSB (solar system barycentre) the discovery and return events are (CT, long, lat):

1846 Sep 23 23:06:47 329-05-33.323 -00-31-35.195
2011 Jul 11 21:50:34 329-05-33.323 -00-31-36.687

Relative to the SUN the corresponding events are (CT, long, lat):

1846 Sep 23 23:06:47 329-06-10.844 -00-31-35.998
2011 Jul 12 21:34:05 329-06-10.844 -00-31-37.197

So you can see there's almost a day's difference between using the SSB and the SUN. This is consistent with the estimates produced by others, if not in detail.

Also note that I've used CT not UT. CT is a uniform time scale used by the JPL ephemerides, whereas UT is non-uniform because of its connection with the Earth's rotation. More importantly there is a significant difference between the time scales, currently about 66 seconds and growing, although at the time of discovery it was only 7 seconds. I suspect that your results for the return time are strictly CT, so are out by more than a minute, though still well within your "tolerance" of 15 minutes.

By the way, CT has nothing to do with considerations regarding light-times. It is just the independent variable used in the numerical integrations that produce the JPL ephemerides (a time coordinate to complement the space coordinates). It is thought to be a uniform time scale and corresponds closely with other astronomical time scales you might have heard of: TDB and TT.

These results were obtained from JPL's DE405 ephemerides (as used by Horizons) but I have also done the calculations with a more up-to-date version, DE423, with the following results:

Relative to the SSB (CT, long, lat):

1846 Sep 23 23:06:47 329-05-32.032 -00-31-35.209
2011 Jul 11 20:23:26 329-05-32.032 -00-31-36.702

Relative to the SUN (CT, long, lat):

1846 Sep 23 23:06:47 329-06-09.552 -00-31-36.012
2011 Jul 12 20:06:58 329-06-09.552 -00-31-37.211

Here the return events are about 1h 27m earlier than those obtained from DE405, although the events still occur on the same day. This might give you some idea about the errors involved in using DE405.


jir1667 said...

So we can see that the major issue is with the choice between using the SSB or the SUN as the centre of motion. Like you I prefer the SSB because the motion of Neptune more closely follows a keplerian path than it does around the SUN. As you have shown the heliocentric orbit has false perihelia and aphelia, which are peculiar to Neptune and are a direct result of the intrinsic motion of the SUN. The SUN's motion complicates things and should really be excluded.

But there is another reason.

We can use the ecliptic latitude to estimate the angular difference between the events. This gives 1.5 and 1.2 arcseconds, from the SSB and SUN respectively, the former confirming your own result. This might suggest that the discovery and return events will be seen to overlap, which is certainly the case for the SSB (as you have depicted) but from the SUN it isn't. Again, this is purely an effect of the SUN's motion, a parallax effect in fact as the SUN is in a different position between the events. From the SUN the angular difference is about 21.5 arcseconds (DE405), mostly in longitude.

You can look at it in another way. From the SSB the discovery and return events could be termed "sidereal" whereas from the SUN they would be "synodic", and it is the former which better reflects the orbital period of Neptune. The synodic period looks to be about one day longer than the sidereal period, accounting for the excess of 21.5 arseconds in longitude.

Interestingly from the SUN, at about 2011 Jul 11 21:50 CT (DE405) there is an overlapping conjunction between Neptune and its discovery direction. This is very close to the return time as seen from the SSB. I don't quite understand this but I suspect it is because at this time the SUN happens to lie close to a line between the SSB and the discovery/return positions and is just a coincidence. Perhaps worthy of closer scrutiny though...


jir1667 said...

Further to my last observation about conjunctions, I had the following scenario in mind.

First determine the 3D inertial position of Neptune at the time of its discovery and mark it with an imaginary flag. Neptune then continues on its merry way around the solar system and we would like to know when it passes by the flag again. As one way of doing this we can find the time when the angular separation between the flag and the returning Neptune is at a minimum from whichever vantage point we choose, in our case either the SSB or the SUN.

The results are shown below. I've done the calculations using DE405 and DE423, as before, but I've also included IMCCE's INPOP10a as a modern independent source of planetary ephemerides which is comparable to JPL's latest data for accuracy.

Times are given in UT this time (properly taking account of CT-UT) and I also show the minimum angular separation attained in each case.

DE405 (UT, arcseconds)
From SSB: 2011 Jul 11 21:46:31 1.491
From SUN: 2011 Jul 11 21:46:26 1.491

DE423 (UT, arcseconds)
From SSB: 2011 Jul 11 20:19:22 1.491
From SUN: 2011 Jul 11 20:19:17 1.493

INPOP10a (UT, arcseconds)
From SSB: 2011 Jul 11 20:03:41 1.492
From SUN: 2011 Jul 11 20:03:37 1.491

The obvious thing to note is that the time of conjunction is virtually the same whether we use the SSB or SUN as the centre of motion.

You can also see that the times determined from DE423 and INPOP10a agree quite well with each other compared to those from DE405. Again, this can be viewed as an improvement in the accuracy of the more up-to-date ephemerides.


jir1667 said...

Also, in all cases the angular separation is such that we could call the conjunction "overlapping", as shown for example in the image at the top of this post. In fact the view from the SUN is pretty much the same as seen from the SSB (though not exactly in the same part of the sky, relative to background stars).

[A clarification about using words like "view" and "seen" is appropriate here. I do not mean what the event will actually look like if I could stand at the SSB or the centre of the SUN and peer through a telescope at the distant planet. Such a view would be light-time retarded, giving us a look at Neptune (and flag) as it would have appeared some hours before. However, here we are mostly interested in the physical geometric motion of Neptune in the gravitational field of the solar system, in which light-time effects are not considered. That is, where Neptune actually is rather than where it appears to be by a distant observer. Alternatively, pretend the speed of light is infinite!]

But of course the most important observation, considering that what we really want to know is when Neptune's "birthday" will be, is that the conjunction occurs on the same day whichever centre or ephemeris we use.

So using the results from DE423 and INPOP10a I would estimate that Neptune will have completed one orbit on 2011 June 11 at 20:12 UT +/- 16m.

Clearly, and probably, not the final word on the matter. As someone said, it's a matter of definition. But a definition, like the one I have used here, that can give a return time from the itinerant SUN consistent with one from the rock-steady SSB is worthy of consideration.


Bob said...

Fabulous stuff, John!

I want to have a good look through the data you've presented before I say anything else, but it looks sound and it seems to be a step ahead of what I presented in the post. I'll add edits into the main post to draw attention to the updates or corrections that you've highlighted in your comments.

Thanks for your input. I agree with you - we have all this understanding and precision data on the motion of the planets: why not delight in making use of it to explore the subtleties of their orbits, to weigh up different interpretations and to get results that are as accurate as we're able to.

I'll come back to look at this properly when I get the chance :)

jir1667 said...

Of course, what I should have written is:

"...Neptune will have completed one orbit on 2011 JULY 11 at 20:12 UT +/- 16m."

Time for a break!


Maarten said...

I know it is a bit late, but in this period Neptune is still in the same spot in the sky as it was in 1846. I am making a documentary about the topic of the discovery and here is an edit I did for the anniversity in July, with a small part of the material: Searching for Neptune, to be found on

Bob said...

That looks like a great little film, Maarten - thanks for posting it.
Good luck with your project.

Post a comment

If you have questions not relating to this post, you can email me.