Lunar Nodes & Eclipse Seasons

The two doorways for an eclipse

If the Moon orbited in exactly the same plane as Earth orbits the Sun, we would get a solar eclipse at every new moon and a lunar eclipse at every full moon — a pair of eclipses every single month. We don't. Instead eclipses are rare, and when they come they come in pairs or trios bunched into a few weeks, twice a year. The reason is a 5° tilt.

The Moon's orbit is inclined about 5.14° to the ecliptic — the plane of Earth's orbit, which is also the apparent yearly path of the Sun across the sky. A tilted circle and a flat circle that share a centre intersect along a line, and that line pierces the Moon's orbit at exactly two points. Those two points are the lunar nodes: the ascending node, where the Moon climbs north through the ecliptic, and the descending node, where it sinks south. They are the only two places where the Moon is actually in Earth's orbital plane.

For an eclipse, three things have to be in a straight line: Sun, Earth, Moon. They can only be in a line if the Moon is on the ecliptic — i.e. at a node — at the same moment the alignment occurs. So eclipses happen only when a new or full moon coincides with the Moon being near a node, and that in turn only happens when the Sun's direction in the sky points close to the line of nodes. Those alignment windows are the eclipse seasons.

The geometry: tilt, nodes, and the line of nodes

Picture two hoops sharing a centre at Earth. One is the ecliptic. The other, tipped 5.14°, is the Moon's orbit. They cross at two diametrically opposite points. The line through Earth connecting those crossings is the line of nodes. The half of the Moon's orbit on one side of the ecliptic is north of it; the other half is south.

The Moon's distance above or below the ecliptic — its ecliptic latitude β — varies sinusoidally as it goes around. At the nodes β = 0; a quarter-orbit away, β reaches its maximum of about ±5.14°. To first order,

β(t) ≈ i · sin(2π t / T_drac)        i = 5.14°,  T_drac = 27.212 days

where t is measured from a nodal crossing and T_drac is the draconic month. Five degrees of latitude on the sky is ten full Moon-diameters, so for most of the month the Moon is far too high or low to line up with anything. Only within a few degrees of a node is β small enough for the shadows to reach.

Crucially, the Sun also lives on the ecliptic by definition — its ecliptic latitude is always essentially zero. So the only way to get Sun, Earth and Moon collinear is for the new/full moon to fall while the Moon is near a node. The Sun reaches the longitude of a node twice per year (once at each end of the line of nodes), and those two passages bracket the two eclipse seasons.

Ecliptic limits: how close is close enough

"Near a node" can be made quantitative. An eclipse requires the geometric separation between the Moon and the Sun-Earth axis to be smaller than the combined angular radii of the bodies and shadows involved. The threshold is expressed as an ecliptic limit: the maximum solar elongation from the node for which an eclipse is still possible.

Limit type	Eclipse	Angular limit from node	Outcome
Major solar limit	Solar (partial certain)	±15.4° – 18.5°	An eclipse must occur
Minor solar limit	Solar (possible)	±9.9° – 11.8°	An eclipse may occur
Major lunar limit	Lunar (penumbral certain)	±9.5° – 12.2°	An eclipse must occur
Minor lunar limit	Lunar (possible)	±3.8° – 6.0°	An eclipse may occur

The limits are ranges rather than single numbers because the Moon's and Sun's distances — and therefore their apparent sizes and the geometry of the umbra — change over the eccentric orbits. The solar limits are wider than the lunar limits because the Sun's apparent disk plus the Moon's gives a larger target than the relatively narrow cross-section of Earth's umbral shadow at the Moon's distance. This is the deep reason that solar eclipses are slightly more frequent globally than lunar eclipses, even though any given location sees more lunar eclipses (a lunar eclipse is visible from the entire night hemisphere, a total solar eclipse only along a ~100 km track).

Why a season always lands at least one eclipse

Convert the solar ecliptic limit into time. The Sun moves along the ecliptic at about 0.985° per day, so crossing a major-limit window of roughly ±17° takes

Δt = 2 × 17° / (0.985°/day) ≈ 34.5 days

So a solar eclipse season runs about 34–37 days. The synodic month — new moon to new moon — is 29.53 days, which is shorter than the season. That single inequality guarantees the result: any window 34 days wide must contain at least one new moon, so every eclipse season contains at least one solar eclipse. When the season is at its longest and a new moon falls right at the start, a second new moon can squeeze in before the season ends, giving two solar eclipses in one season (one of them always partial, near the edge of the limit).

Lunar eclipses have the narrower ±10° limit, a window of only about 20 days — shorter than the 29.53-day synodic month — so a given season is not guaranteed to contain a full moon inside the limit. That is why some eclipse seasons produce two solar eclipses and no central lunar one, while others give the familiar solar–lunar pair.

Nodal regression and the eclipse year

The line of nodes is not fixed in space. The Sun's gravity exerts a torque on the Moon's tilted orbit — the same physics that makes a tilted spinning top precess — and the result is that the nodes regress, sliding westward (retrograde) around the ecliptic. One full circuit takes

T_node = 18.6129 years   (nodes move ≈ −19.34° per year, retrograde)

Because the nodes are drifting toward the oncoming Sun, the Sun returns to the same node sooner than it completes a full sidereal lap. The interval between successive passages of the Sun through the same node is the eclipse year:

1 / T_eclipse_year = 1 / T_tropical_year + 1 / T_node
T_eclipse_year = 346.620 days   (about 18.6 days shorter than a calendar year)

Half an eclipse year — 173.31 days — separates the two eclipse seasons. Since that is shorter than half a calendar year, the seasons creep about 19 days earlier on the Gregorian calendar each year. Track eclipses over a decade and you watch the seasons march steadily backward through the months; after about 9.3 years they have moved half a year, and after the full 18.61-year nodal cycle they return to roughly the same dates.

The cadence: 4 to 7 eclipses a year

Two eclipse seasons per eclipse year (346.62 days), each guaranteeing at least one solar eclipse and usually delivering a solar–lunar pair, sets the rhythm. Every calendar year has a minimum of four eclipses (two solar, two lunar) and a maximum of seven. Because the eclipse year is shorter than the calendar year, a third partial eclipse season can occasionally squeeze into a single Gregorian year — that is how rare seven-eclipse years (like 1935) arise.

Quantity	Value	Note
Synodic month (new→new)	29.5306 days	Phase cycle
Draconic month (node→node)	27.2122 days	Sets eclipse possibility
Anomalistic month (perigee→perigee)	27.5546 days	Sets eclipse magnitude
Eclipse season length (solar)	≈ 34–37 days	From ±17° major limit
Half eclipse year (season spacing)	173.31 days	Between the two seasons
Eclipses per calendar year	4 (min) – 7 (max)	Usually 4 or 5
Saros period	6,585.3211 days (≈ 18 yr 11.3 d)	Geometry near-repeats

The Saros: three months that all line up

The genius of the Saros cycle is that three different lunar months come back into step at once. After 223 synodic months,

223 × 29.5306 d = 6585.32 d
242 × 27.2122 d = 6585.36 d   (draconic — Moon back at the same node)
239 × 27.5546 d = 6585.55 d   (anomalistic — Moon back at same distance)

All three equal about 6,585.3 days — 18 years, 11 days, and 8 hours — to within fractions of a day. So after one Saros the Sun is back near the same node (eclipse possible), the Moon is back at the same phase (new or full), and the Moon is at nearly the same distance (so the eclipse has nearly the same character — total, annular or partial). The result is a Saros series: a family of eclipses, each one Saros apart, that evolves slowly over 12–15 centuries. The extra ⅓ day shifts each successive eclipse about 120° westward in longitude, so it takes three Saroses — a triple Saros or exeligmos, 54 years and 33 days — to bring an eclipse back to roughly the same part of Earth.

The Babylonians recognised the 18-year rhythm by the 7th century BCE and used it to predict eclipses without any model of the underlying geometry — pattern recognition standing in for celestial mechanics. The word "Saros" was attached to it by Edmond Halley in 1691, borrowing a Babylonian term.

Numbers worth remembering

• Tilt: the Moon's orbital inclination to the ecliptic averages 5.145°, oscillating between about 5.00° and 5.30° over roughly 173 days due to solar perturbations.
• Apparent sizes: the Sun and Moon are each about 0.5° wide. A 5° tilt is ten lunar diameters — which is why the Moon usually clears the Sun comfortably at new moon.
• Node motion: 19.34° per year retrograde, i.e. about 3.2 arcminutes per day; a complete circuit in 6,798.4 days (18.61 years).
• Standstills: when the regressing node lines up so the Moon's 5.14° tilt adds to Earth's 23.44° axial tilt, the Moon reaches declinations of ±28.6° — the major lunar standstill, last in 2006 and 2024–2025, next around 2043.
• Origin of the tilt: the 5° inclination is a fossil of the giant-impact formation of the Moon and subsequent tidal evolution; it is measured relative to the ecliptic, not Earth's equator, because solar tides dominate the precession at the Moon's distance.

Where it shows up

• Predicting any eclipse. Every modern eclipse prediction starts from the position of the Sun relative to the lunar nodes. NASA's eclipse canon tags each eclipse with its Saros number — e.g. the 2017 "Great American Eclipse" was Saros 145, member 22 of 77.
• The 18.6-year tidal cycle. Nodal regression modulates the lunar contribution to ocean tides with an 18.61-year period, the lunar nodal tide, detectable in long tide-gauge and sea-level records and relevant to coastal flooding statistics.
• Major lunar standstills and archaeoastronomy. Sites such as Callanish and possibly Stonehenge appear aligned to the extreme moonrise/moonset azimuths that occur at the 18.6-year standstill — the same nodal cycle, expressed on the horizon.
• Spacecraft and satellite orbits. The very same Sun/Earth-oblateness torque that regresses the lunar nodes regresses the nodes of artificial satellites; choosing an orbit where this nodal precession equals 0.9856°/day yields a Sun-synchronous orbit, the workhorse geometry of Earth-observation satellites.
• Historical chronology. Babylonian Saros tablets and recorded eclipses are used by historians to pin absolute dates, because the regressing nodes make each eclipse's date, path and Saros membership a near-unique fingerprint.

Common misconceptions and edge cases

• "Eclipses happen at random." They are among the most predictable events in nature. The constraint is purely geometric: Sun near a node, Moon at new (solar) or full (lunar). Everything else follows from the three lunar months.
• "The nodes are physical objects." A node is just a geometric crossing point — the intersection of two planes. There is nothing there. What matters is the Moon's latitude being near zero at the right phase.
• "Eclipse seasons are six months apart." They are half an eclipse year apart — 173.3 days — not half a calendar year. The 19-day annual drift is the visible signature of nodal regression.
• "A Saros eclipse hits the same place." One Saros later the eclipse recurs but is shifted ~120° west in longitude and about 0.5° in latitude. It takes three Saroses (the exeligmos, 54 years) to return near the same region, and the path drifts steadily across latitudes over the series' lifetime.
• "The tilt is to Earth's equator." The 5.14° is measured to the ecliptic. Relative to Earth's equator the Moon's orbit varies between 18.3° and 28.6° as the node regresses — which is exactly what produces the lunar standstills.
• "Annular vs total is about the nodes." Whether a central solar eclipse is total or annular is set by the anomalistic month (Moon's distance), not the draconic one. The nodes decide whether an eclipse happens; the distance decides what kind.