Discovery of Neptune: historical notes
Historical notes about discovery of Neptune
The first theory of Neptune’s existence was made by Galileo Galilei, on December 28, 1612, probably one of the most well known astronomers, was once looking at the night sky, searching for new planets. That night, he found something very interesting, it was what seemed to be a star that didn’t twinkle, and therefore Galileo thought he had found a planet! He looked at again a year later to figure out that it was starting to go retrograde and then proved that it was indeed a planet, he wrote it in his drawings of the solar system with no name and showed it to the public, astronomers kept looking for Neptune and eventually found out that it was just a star that was in conjunction with Jupiter witch is what made it look like it was moving to Gallileo, and his small telescope, and was never fought of again until 1821.
In 1821, astronomer Alexis Bouvard was observing Uranus, he kept track of it for months and wrote on his tables that it was constantly getting perturbed by what could only be some unknown body, causing him to believe the existence of Neptune.
In 1844, john couch Adams was fascinated by Neptune’s mysteries, and wanted to prove it’s existence, he gathered a lot of data on the subject by himself, but mainly talked to other famous astronomers to prove Neptune’s existence until 1846.
The final time Neptune was searched for was when Urbain Le Verrier started researching it, unaware of John’s observations, he made calculations of where Neptune would be by finding Uranus, and watching where it was being pulled, and how much via newton’s laws, then he went to make a note to astronomer Johann Gottfried Galle that said to use the refactor at the Berlin observatory and to use his calculations to find Neptune’s position, when Galle got the note he looked for Neptune in the night sky and found it! 1 Degree of where Le Verrier fought it would be.
I would like to point out some statements that I belief are erroneous here, in your historical summary. Just for any future reader to take these into account (don’t hate me please).
This discovery (of Galileo’s observations) was published in Nature by Charles T. Kowal & Stillman Drake in 1980. The article is interesting on its own if anyone wants to research on the history of “the precovery of Neptune”. It seems that Galileo pointed his telescope to Jupiter in a very appropriate moment; our current high-accuracy ephemerids show that Neptune was been occulted by Jupiter in January 1613 (a one in a century event at best). That means that the months before and after that event Neptune was close to Jupiter in the sky and any curious eye trying to learn something about Jupiter would have been able to possibly spot Neptune.
The diagram above shows Jupiter and the relative position of Neptune from 12/28 (December 28th) to 1/30 (January 30th). The second lucky thing is that Jupiter started in January 1613 it’s retrograde motion (apparent from Earth). This is why in the diagram (where Jupiter is fixed) Neptune appears to return. The fact Neptune was slowing down and returning close to Jupiter’s position is also good since that incremented the chances of Galileo to spot it.
Galileo is known to have been able to spot even magnitude 9 stars. Since Neptune never gets above magnitude 8 that means that Galileo indeed should have been able to see Neptune in some observation that moth. The field in the sky where the first observation was made is depleted of bright stars so this “fixa” must have been Neptune indeed.
Let’s look at the actual notes done by Galileo.
Lets zoom in to the upper part of the second page of his notebook:
In the upper part he writes the time when the observation was made. It is December 27th, 1612, 15 hours and 46 minutes after noon. Galileo’s way of formating time seems strange to us now. 15 hours and 46 minutes after noon is just 03:46 am of the next day. So the observation was made at 03:46 local time in December 28th of 1612. The first drawing shows Jupiter and 3 moons aligned in the same plane, he also puts “distance” measurements between Jupiter and each moon; the one in the left was at 9 Jupiter radii, and the two on the right where at 9 and 10,3 Jupiter radii respectively. He wrote “fixa” for a star in the upper left part of the image which he would take as a fixed point during this set of Jupiter observations.
The lower drawing shows the situation a few hours after that. The moons have moved a little (one of the four moons has just started to emerge from the glare of Jupiter’s disk) and “fixa” still remains more or less fixed in place.
Now compare the first drawing with the actual positions calculated using modern ephemerids:
As you can see “fixa” must be the planet Neptune and not a star. The moons marked by Galileo in the first drawing are (from left to right) Ganymede, Europa and Callisto.
Now let’s jump one month to these notes:
Here again we see Jupiter and three moons (from left to right: Ganymede, Europa and Callisto at 5,10 , 8,40 and 20,40 Jupiter radii away from Jupiter according to Galileo). This observation was made the 28th of January 1613, 6 hours after sunset. A star labeled with an “a” can be seen 29 Jupiter radii away in the far left of the image. This new fixed point he used was indeed a star. The star is SAO 119234 (or HD 105374). He used these fixed points as references to understand the motion of Jupiter in the sky. Little he knew that the fixed point used a month before, “fixa”, was Neptune and therefore not fixed at all. He wanted to use a secondary fixed point for this new observation so he took another star, “b” which was a little farther away than “a” but basically in the same direction. Since “b” couldn’t fit in the space left in that page he splitted the drawing and continued the Jupiter-a-b line in the lower right part of the image. Well, it turns out that “b” was “fixa” again that had displaced in the sky (look at the first image of this post). “b” was Neptune. You can see how well it agrees with our ephemerids of that observation here:
The interesting thing here is that in this observation Galileo wrote that “b”, which was also observed in the preceding night, and “a” seemed to be more remote from each other (inter se) than the previous night. This means Galileo was able to correctly detect the motion of Neptune, even if he dismissed the idea and never thought of it again. I want to emphasize “correctly detected” because he was aware that Jupiter and “b” would be moving relative to one another but the fact that “b” separated from “a” (both supposedly fixed celestial marks) was really unexpected.
Le Verrier predicted the mass and orbital parameters of the planet (as well as the current position) so he had a very good estimate as to how massive and far away Neptune was. That said it is true that semi-major axis and mass were educated guesstimates.
By the way it is also worth noting that Adams is currently disregarded as a contributing force for the discovery of Neptune. Nearly all the credit goes to Urbain Le Verrier.
What I’m getting at is if he simply observed Uranus for a few months and tried to observe how much it got pulled by Neptune during that time, then the effect is too small to be detected with the telescopes available in the 1800s.
Let’s suppose Uranus and Neptune were as close as they could possibly get to each other (about 10AU apart). The acceleration of Uranus due to Neptune is
which during a time Δt will shift Uranus’ position by
If this shift were oriented in the best possible way relative to Earth (perpendicular to our line of sight), then the change in Uranus’ position on the sky during this time is
Let’s plug in the numbers. Neptune has a mass of M = 1.02×1026 kg. Let the observing time be 6 months. Then the shift in Uranus’ position due to Neptune pulling on it during those 6 months, assuming they were only 10 AU apart, is 370 km. From our vantage point about 20 AU away, this can produce, at most, an angular shift of 0.025 arcsec.
Historically, the largest telescopes available in the 1800s were less than 1.8m in aperture. (The biggest at 1.8 m was the Leviathan of Parsontown, which was not complete until 1845, after the irregularities in Uranus’ orbit had already been noticed). The minimum angle that can be resolved with a 1.8 m optical telescope is 0.07 arcsec. (Even this 0.07 arcsec is still unobtainable because atmospheric turbulence limits resolution as well, and rarely is better than 0.5 arcseconds. Today we overcome this with either space-based telescopes or adaptive optics.)
So in the 19th century, no telescope in the world could possibly hope to see Uranus be perturbed by Neptune over the course of a few months. There has to be more to the explanation of how Neptune’s influence was observable.
In fact, there is. Try checking the wikipedia article on Neptune’s discovery for more information on the perturbations. 🙂
That’s actually quite interesting and I didn’t realized at all how minuscule the effect should be. Even with a simple model you have show how tiny this is. As I see, in reality it would be even worse since that acceleration could be archived only for short periods of time. After and before the actual opposition, Neptune would be farther away and we know that gravitational force is fairly sensitive to distance.
Also in the opposition of Uranus and Neptune there is no place on Earth’s orbit to have a reasonable angle to actually see the displacement (produced when the perturbing influence is at max). Since there’s no situation where Earth could be perpendicular to the line between Uranus and Neptune (at their closest approach) the angle subtended by Earth’s orbit at Uranus distance has to be taken into account. I’ve calculated the real displacement to be around 0.052 times the one calculated by Watsisname when we incorporate this to the model. So instead of 370 km of displacement, from the best vantage point in Earth’s orbit, we would at most see 19 km!! Even with current technology it would be nearly impossible to notice at all (if we had 19 km resolution we would be mapping Uranus’ moons from Earth).
I think the answer here has to do with two things (but only one of them seems to be relevant in the end);
First: a larger observational baseline is needed (not just a few months).
Second: Neptune’s tug, even if weaker, should be more constant and oriented tangential to Uranus’ orbit when they are still far from opposition.
The first thing is easy. Le Verrier used only observations made between 1781 and 1821. That’s 40 years of observations. Let’s take 8 years for example as a baseline. Uranus and Neptune are close to their opposition all this time so we can still use the approximation of the gravitational tug staying constant. The good thing of the formulas provided by Watsisname is that time is squared so if we increase a little the observational baseline we have a significant increase in the effect. In this case and even considering my 0,052 factor we get 0,34 arcseconds of separation between the real Uranus and the “predicted” Uranus after 8 years. Barely noticeable still.
So the actual perturbing effect is, maybe (I don’t know so please tell me if you have better insights) due to the second option. The actual configuration where the Sun-Uranus-Neptune angle is 90º (in this configurations Neptune’s tug is directed tangentially to Uranus’ orbit) was archived around the year 1800 and again around 1850 (in the middle there was the opposition). In this configuration Uranus and Neptune were separated by a distance of r = 25,4 AU. But also in this configuration we (the Earth) can be at a good angle to see Uranus’ displacement (perpendicular to the Neptune-Uranus line if you want), so here there’s no need for my 5,2% reduction of the effect. You can admire the 100% of the displacement in this configuration. The problem is that the tug is weaker since Uranus and Neptune are farther apart than when they are at opposition. But still considering just 6 years it would mean a 0,59 arcseconds discrepancy (surpassing 1 arcsecond after 8 years only), something that starts to be noticeable I guess.
Here we have a graph (taken from Le Verrier: Magnificent and Detestable Astronomer) that shows the “residuals” of Uranus’ longitude across 150 years or so.
O stands for the observed longitude of Uranus and C stands for the computed longitude of Uranus after taking into account the perturbations of all the planets. If there was no Neptune this should have been a straight horizontal line at 0 arcseconds, since the computed position of Uranus and the observed one should agree, so the residual (which is the difference between both values) should be zero. But Neptune exists and the residual fluctuates.
Now lets see the actual configuration of the planets through the XIX century (this has been taken from “The Case of the Pilfered Planet” an article published in the Scientific American journal)
So lets compare both images. Around 1800 and 1850 we have the configuration I mentioned (where Neptune’s gravitational pull seems to be tangent to Uranus’ orbit). Around 1820 we have the opposition. As you can see Uranus’ predicted position is lagging behind with respect to Uranus’ actual position because Neptune is accelerating Uranus along his track (his orbit). The opposite can be seen in 1850 when the predicted position of Uranus is ahead of the actual position (because Neptune is pulling Uranus in the opposite way so it decelerates). This is also evident in the residuals graph; in 1800 the computed longitude for Uranus was larger than the observed one so the residuals where negative, but close to 1850 the residuals are positive since the observed longitude is larger than the computed one.
Also interesting to note the fact that at opposition the effect is the lowest possible, even if paradoxically the gravitational tug is the greatest, because of the proyection effect of the acceleration I’ve talked about in my second statement. You can see that the residuals disappear at opposition and Uranus’ position does not differ from the predicted one (also, bare in mind that the separation in the second image is greatly exaggerated to make it clearer).
But I still miss a more in depth explanation. I’ve shown that in 1800 and 1850 the effect is more noticeable than in 1820 when the planets are closest to each other. But still we get around 1 arcsecond displacements after 8 years. Too little in my opinion. Also in the residuals graph they are using one or two orders of magnitude above that (tens or even near a thousand arcseconds discrepancy) for the residuals so probably I am completely lost in terms of what the residuals actually are and how they are actually calculated here. I can only make little sense of the overall trend in the different years. Also I can’t understand why the graph of the residuals behaves like that, it seems to increase a lot when the planets are farther apart for some reason and it is not sinusoidal as I intuitively expected. Also the graph seems to indicate that the zero point for the residuals (where the difference between observation and prediction is zero because the prediction was made from this moment) was taken around 1775, right? Why? And why it is so close to the expected time on which Uranus and Neptune should be the farthest apart possible? Coincidence? Those are the kind of things that make me think that I haven’t understood the basics of the ideas behind the discovery. I would like to know more. a lot more. It would be crude but beautiful to see the full demonstration just to feel the power of Newtonian vision.
But sadly this is a difficult topic. Jean-Babtiste Biot attempted to explain Le Verrier’s methods in six papers he published in 1846 (just before the discovery) and in the third paper he wrote “As I progress in the task I have undertaken, the difficulty of the subject seems to increase”. I feel the same. The more you know the more complex it seems to be. I hope there’s someone here that has the patience to explain it in detail because it is very exciting subject in my opinion and there should be some place in the Internet that has an explanation like that.
SOME NEWS: I just found a web where there is an interesting explanation of Le Verrier’s calculations but it is in French and I’m unable to understand it. If some French fellow forum members want to read it please show the rest of us what is all about 😀
It’s amazing what we can do with computers today. I did try directly computing orbits with and without perturbation just to try to get some intuition for how the perturbation effect works. For starters I assumed co-planar, circular orbits, and also that only Neptune perturbs Uranus and not vice versa (which probably matters very little). For the initial condition I put Uranus directly opposite Neptune, and then iterated through 1 hour time steps. I add vertical lines to show when they are closest together (by direct distance, not angle):
I think this region (zoomed in between a conjunction and following opposition) fits with Le Verrier’s figure:
There are still significant differences between them (like the amplitude, and width of the minimum vs. maximum), which I guess is mostly due to Uranus’ eccentricity. I must factor that in to ‘perturbation simulator v0.2’. The shape and overall trend of the curve also depends a lot on the initial position of Uranus relative to Neptune. I have no idea what Le Verrier used in detail. But any rate I think the physics operating here is pretty clear like you said. The variation occurs because while Uranus is approaching Neptune, Neptune causes it to accelerate, and then decelerate after closest approach. It’s a very slow effect since the gravitational attraction is weak when they’re far apart, but it has a lot of time to accumulate.
I think this is also exactly the explanation behind exoplanet transit timing variations. We see the projection of the planet in its orbit against its star, so these same perturbations will cause transits to appear earlier or later than they otherwise would if there were not additional planets in the system. I’m not sure of the exact details for how we go from the timing observations to predicting where the other planets are, but it must be essentially the same problem, solvable by computer with Monte Carlo methods.
I’m still very curious in how exactly they solved Neptune’s orbit without computers. For a computer this is a problem that takes a few seconds. But by hand…? I can’t even imagine. Plus they were observing the effect, and had to deduce the cause, by successively adjusting the “guess” for the location of Neptune to get closer and closer to what was observed. This is more in the realm of “mathematical methods of physics”, but a very interesting subject.