An eclipse of the Sun occurs when the new Moon sits between the Sun and the Earth, thus blocking the light coming from the Sun. Because of the incredible coincidence that the apparent size of both these heavenly bodies is currently almost precisely the same, the obscuration is exact, with the Moon just covering the entire disc of the Sun to produce the spectacular sight of the total eclipse as we know it.
However, several factors conspire to upset this simple picture (else we would have a perfect eclipse at every new Moon!). Firstly, the orbits of both the Moon around the Earth and the Earth around the Sun are not exactly circular and so the apparent size of both the Sun and the Moon as seen from the Earth do vary. This results in the Moon sometimes not being large enough to completely cover the Sun, so a thin "halo" of light is left round the edge to give the so-called "ring of fire" effect. This is called an annular eclipse and is much less spectacular than a full eclipse. In fact, this sort of eclipse will become more common as millennia pass because the Moon is actually getting further away from the Earth (by about 3.8cm per year!), and thus apparently smaller. This is because the action of raising tides on the Earth takes energy from the Moon's motion, slowing it down in its orbit. To maintain a stable orbit at this steadily slowing speed requires the Moon to orbit further and further from the Earth, resulting in it slowly drifting away. It is estimated that periods when full eclipses are not possible will happen from about 620 million years from now with the last full eclipse happening in around 1210 million years time - not a problem to be worrying about for a while, therefore!
The second complication is that the Moon's orbit round the Earth is not exactly in the plane of the Earth's orbit round the Sun - it's tilted by about 5 degrees in fact. This means that for most new Moons, as seen from the Sun the Moon is either slightly above the Earth or slightly below, and so there's no eclipse. However, if a new Moon should occur when the Moon is at one of the two points in its orbit when it is passing through the plane of the Earth's orbit (called "nodes") then it will indeed be directly between the Sun and the Earth, resulting in an eclipse. Time for the first diagrams, I think!
Below, we have the Moon orbiting the Earth which is in turn orbiting the Sun - if we assume the viewpoint is from above the Sun's north pole then both bodies orbit their parent in an anticlockwise direction. As the Moon's orbit is tilted with respect to the plane of the Earth's orbit, part of it (the light-green bit) will be above the plane and part (the dark-green dashed bit) will be below. The nodes, the points where the Moon's orbit cuts the plane of the Earth's orbit, are labelled A (for ascending) and D (for descending): so called because the Moon will be rising above the plane at the ascending node and dropping below it at the descending node. The black line joining them is called the "line of nodes". The smaller diagram gives the view from the Sun along the Sun-Earth line, and is to scale with the view from above the Sun's pole. It is clear in this view how half the Moon's orbit is above the plane and half below it.
| |
| When the line of nodes is not aligned with the Sun-Earth line, as here, it is clear that the Moon will pass below the Earth when it moves between the Earth and the Sun i.e. at New Moon. There can thus be no eclipse. The same conclusion would hold if the Moon had just passed the ascending node, only this time the Moon would be above the Earth when New. | Only when the line of nodes points along the Sun-Earth line will the Moon pass directly between the Earth and the Sun at New Moon, producing an eclipse (labelled E and shown in pink). |
However, even when everything has fallen just right for a solar eclipse to happen it will not be visible from the entire daytime side of Earth. During an eclipse, the spot where the Moon's full shadow falls on the Earth is usually barely 100 miles wide and so only from within the corridor swept out by this spot as it races across the Earth will the eclipse be visible as total. In a further area on both sides of the band of totality it will be seen as partial, but from anywhere outside this area there will be no eclipse visible at all. This explains why "eclipse chasers" must take great care to ensure they are in the right place - thank goodness for GPS! It makes a jolly good excuse for holidays in far-off places though, as you will see from my descriptions!
Eclipses do not happen at random intervals, because of the need for a new Moon to coincide with a nodal crossing. As previously mentioned, there are two nodes and so I think it should be clear from the diagrams above that the coincidence period will be about 6 months. Now it just so happens that six times the interval between new Moons [29.53 days x 6 = 177.18] is very nearly equal to six and a half times the interval between node crossings [27.212 days x 61/2 = 176.88] - the extra half is needed because the Earth will be on the other side of the Sun after 177 days and so the Moon must do another half orbit to be on the correct side of the Earth i.e. at the node which is between the Earth and the Sun. In most cases, therefore, successive eclipses are separated by an interval of 177 days. This period is known as an eclipse season.
This is not the end of the story though, as the Moon also needs to lie on the Earth-Sun line for there actually to be an eclipse which, as we have seen, would seem to happen 6-monthly. At face value, because 177 days isn't equal to 6 [calendar] months you would think eclipses should thus be very unusual. However, the Sun-Moon-Earth alignment doesn't have to be absolutely precise because the Earth is large enough that the Moon's shadow will still strike it even if the alignment isn't exact, and fortunately there is an additional complicating factor (there always is!), due to the fact that the orientation of the line of nodes is not fixed in space. Because of the way the gravitational attraction of the Sun and the Earth affect the Moon's orbit, the line actually rotates clockwise, making one turn every 18.61 years. This means the nodes re-align themselves with the Earth-Sun line every 173.31 days rather than the 6 months one might imagine. Clearly, 173.31 is closer to 177 than is 182.62 (i.e. 6 months), and in fact it is "near enough" for an eclipse of some sort to occur at every eclipse season. The discrepancy in numbers just means that successive eclipses will drift through a "window of opportunity". Those that are just in the window (i.e. when the Sun-Moon-Earth alignment is "just good enough") will be partial eclipses. As the alignment gets steadily better they will become total or annular but after crossing the middle of the window (the point of best alignment) they will eventually revert to partials again before the whole cycle repeats.
The slight imprecision allowed also means that it is sometimes the case that the alignment is good enough to produce an eclipse either one lunar orbit before the full 177 days is up or just one further orbit afterwards. Successive eclipses can thus be separated by 148 or 29 days as well as 177: there can therefore be more than the two or three eclipses per year one would guess from the eclipse season length. I've relegated the somewhat complex discussion of sequences and the way they change over time to another page - click here if you're feeling in need of a little mental exercise! Suffice it to say that as a result of this complication the number of solar eclipses possible in a given calendar year can vary from a minimum of two to a maximum of five. Five in a year is very uncommon (just 24 cases from 1750 BC to 3400 AD!) and only either the first or the last can be non-partial. The last year with five was 1935 (four partials then an annular), the next will be 2206 (an annular then four partials). Four in a year happens reasonably often though - about 1yr in 10. The last times were in 2011 (4th January, 1st June, 1st July, 25th November), and 2018/9 (15th February, 13th July, 11th August, 6th January - note that this is a case of "4 in a 12month period" rather than "4 in a calendar year"). The next will be in 2029 (14th January, 12th June, 11th July, 5th December). All these eclipses are partials. The next calendar years with four eclipses which are not all partial are 2076 (6th January, 1st June, 1st July, 26th November) and 2094 (16th January, 13th June, 12th July, 7th December), when in each case the first one is total. The next with two non-partials is 2195 (10th February, 7th July, 5th August, 31st December) when the first is annular and the third is total.
The 177-day period mentioned above is just one of many periodicities arising out of the need to reconcile the new Moon, node crossing and node position timings. The most well-known (and historically most important) is the Saros period of 6585.32 days (18years 10/11days [depending on leap years] and 8hrs). It was known about in ancient times and used to predict eclipses, the occurence of which was thought to herald major events. The name doesn't have any historical significance though, having been given by the astronomer Edmund Halley in 1691. The period is equal to 223 new Moon periods and is also very close to 242 node crossing periods (6585.36 days) and 38 "node line-up" periods (6585.78 days). As a further bonus, it is also almost equal to 239 times the period between closest approaches of the Moon to the Earth (6585.54 days). All this means that one Saros period after a given eclipse the Sun, Moon and Earth will be almost exactly back to their same relative positions and so you will get an eclipse that is almost identical to the first. The only real difference will be that, due to the odd third of a day, its timing will be shifted by 8hrs (i.e. the track will have moved round the Earth by 120degrees). To see the Saros in action, observe that the eclipses in 2018/9 listed above are all 18yrs and 10/11/12 days (depending on how the leap days fall and whether the shift in timing takes the eclipse past midnight) later than those in 2000. This shows that not only is the geometry repeated after one Saros period but also often the pattern of eclipses as well.
A series of eclipses each separated by the Saros period is known as a Saros series. Because the above numbers are not exactly equal, eclipses in a given Saros series will slowly shift in time and space (in the same way as successive eclipses move through the window of opportunity) until the alignments are no longer good enough to produce an eclipse. This shift is much slower than for simple sequences of eclipses, however, as the agreement between the various orbit numbers is very much more exact. In fact, it takes about 72 Saros cycles for the alignments to move from "just good enough" through "perfect" to "near miss" - about 1300yrs! Like the 177-day sequences, and for the same reason, eclipses in a given Saros series also start and end with partials, with totals/annulars in the middle of the run. Note that I said "a given Saros series": because eclipses happen much more frequently than once every 18yrs it is clear there must be many Saros series running at the same time, at different stages in their "lifecycle". There are an average of 2.38 eclipses of all types per year so in a Saros period of 18.03 years there will be about 43 eclipses. Only the first and last of these will belong to the same series so on average there must be about 42 series running, in order to account for all the other eclipses. As old series end, new ones begin. If we say that each Saros series has a lifetime of 1300 yrs, to maintain the figure of 42 simultaneous series a new one must begin (and an old one end) about every 31yrs.
The above numbers are, of course, average values and relate to the current epoch. The development of Saros series over time is more complex than for 177-day sequences, mainly because with such a large number of eclipses to play with the scope for variations within a series is that much greater. There are three main effects - periodic changes in the relative numbers of opening partials, totals/annulars, and closing partials; sudden jumps in the total number of eclipses in a series, and a slow decrease in all numbers over timescales of thousands of years. I undertook a major investigation into the reasons for these changes which I have written up in a separate article - click here to read the results. But be warned, it's not for the faint-hearted!
The near-perfect alignments that are required for a solar eclipse are, in fact, also required for a lunar eclipse: it's just that the Moon must be on the other side of the Earth. It thus frequently happens that the alignments that produce a solar eclipse are sufficiently accurate to produce a lunar eclipse about half an orbit earlier or later. For example, in 2000/01 there were total lunar eclipses on 21st January, 16th July and 9th January, each just 15days before or after the solar eclipses. Note that the one on 16th July was 15days after one solar eclipse and 15days before another: a total of three eclipses in just 30days! The calculations for the minimum and maximum number of lunar eclipses in a calendar year give the result of a minimum of zero and a maximum of three (different from the result for solar eclipses due to the different geometry involved). However, it is not possible to have five solar eclipses and three lunar eclipses: the maximum is in fact seven (5+2 or 4+3). The longest possible period without a lunar eclipse is 679 days (1yr 10mths 10days).
