Where these numbers come from

Everything is computed in your browser

There is no server here and no astronomy API. When a page shows you a phase, a distance, a rise time or an eclipse track, your own device has just worked it out. That is why the site keeps working offline once loaded, why this site never sends anything about where you are anywhere, and why a figure here can differ slightly from one you read somewhere else: two implementations of the same astronomy rarely agree to the last digit.

The phase, the distance and the times

Computed from the algorithms in Jean Meeus, Astronomical Algorithms (2nd edition), which is the standard reference these calculations are built on almost everywhere. Phase for a given day is taken at 9 pm your local time — the Moon changes measurably over a day, so “the phase today” needs an hour attached to it, and an evening one is the honest choice for a site about looking at the sky. Positions and the live readouts use the current moment.

Time, and whose day it is

Dates and times are shown in your device’s own time zone, and the day a page calls “today” is your local day. The underlying arithmetic runs in Universal Time and is converted for display. That matters for anything near midnight: a full moon at 00:30 UTC belongs to one date in London and the previous one in Denver, and both are correct. Where a page names a date without a time, it is the local date.

The birthday moon

The birthday explorer takes a Gregorian date from 1 January 1900 to 31 December of the current year. Impossible dates are refused rather than moved: 30 February is an error, and so is 29 February 1900, because 1900 was not a leap year. The weekday is the weekday of the date as entered.

A date is not a moment, and the Moon changes through a day, so one moment is named. Unless you give a time, it is 12:00 UTC on that date, which is still that date almost everywhere on Earth, and the page says so next to the numbers and says how much the lit fraction changed across that UTC day. If you give a local time and a time zone, the moment is that wall-clock time in that zone, converted with your browser’s own time-zone database. A time the clocks skipped is refused; a time they passed twice is put to you as a question rather than guessed.

The numbers come from the same engine as the rest of the site. Checked against JPL Horizons at 23 moments from 1900 to 2026, the lit fraction agrees to within 0.02 percentage points and the distance between the centres of Earth and the Moon to within 50 km. Checked against the US Naval Observatory’s 348 new moons, quarters and full moons in seven years from 1900 to 2026, the times agree to within about three minutes; the engine does not model ΔT, the slowly growing gap between clock time and the smooth time the orbits run on, and that is most of the difference. So those times are shown to the nearest five minutes, and called “about”. The saved reference values are part of the site’s own build checks.

The eight phase names are broad stretches of the cycle, not instants: each of new, first quarter, full and last quarter is given about a day either side of its exact moment, and the crescents and gibbous moons have the rest. A rounded 100% does not by itself mean the exact full moon fell on your date, which is why the page prints when it did. The picture is a rendering of that phase, drawn north up as the Moon looks from the Northern Hemisphere; from the Southern Hemisphere it appears turned round. It is not a reconstruction of that night: weather, the Moon’s colour, whether it was above your horizon and how it was tilted for you are not calculated.

History is matched on the date an event is recorded under, in the Gregorian calendar (an Old Style date is converted and noted). Anything on your exact date comes first, the same day in other years second, and a few events under a similar Moon last, labelled as unrelated to your date. When nothing is on file for the date, the page says so.

The star finder

The star finder works out where selected bright stars and constellations are in your sky. The stars come from the Yale Bright Star Catalogue, 5th revised edition (Hoffleit & Warren 1991), via VizieR/CDS: every star to magnitude 4.5, plus the fainter ones the constellation lines need, 906 in all, kept at the catalogue’s own positions (equinox and epoch J2000.0) and proper motions. Star names are the IAU Working Group on Star Names’ approved names, checked against its catalogue. The constellation lines are this site’s own drawing: the IAU defines constellations as regions of sky and no stick figures, so every drawing is one convention among several.

For each moment the finder carries every star from the catalogue to the date: proper motion, precession (IAU 1976), the main terms of nutation and annual aberration, then altitude and azimuth for your place, from Jean Meeus’s Astronomical Algorithms. Altitudes are geometric: the air lifts things near the horizon by around half a degree, by an amount that depends on the weather, and the finder does not pretend to know it. The Sun and Moon come from the same engine as the rest of the site.

It has been checked against Skyfield, an independent astronomy library, with NASA JPL’s DE421 ephemeris, fed the same catalogue rows: 1,320 star positions at six places from Quito to Longyearbyen and seven dates from 2000 to 2050. Every one agrees to within 38 seconds of arc, about a fiftieth of the full Moon’s width, and the typical difference is about one second. The largest differences are all in 2050, and come from the one thing nobody can know yet: how far Earth’s rotation will have drifted from the clocks by then. The sky map on the same page turns the sky with one rotation for the instant: precession and nutation as above, then the place, applied alike to NASA’s star imagery (Deep Star Maps 2020, NASA/Goddard Space Flight Center Scientific Visualization Studio; Gaia DR2: ESA/Gaia/DPAC) and to the catalogue stars drawn over it, so the two stay together. It leaves out aberration, at most 20 seconds of arc, because a picture of the whole sky cannot be shifted star by star; against Skyfield it agrees to 22 seconds of arc. The Moon and planets come from the site’s planet engine, seen from your place, and agree with Skyfield and DE421 to within a thousandth of a degree at four places in June and December. Its constellation lines are the figures drawn for the IAU by Alan MacRobert of Sky and Telescope as they appear on NASA’s map, traced star to star; the star guides’ own charts keep this site’s simpler drawings of the key shapes. Those IAU charts draw no figure for Microscopium or Mensa, so their lines are Stellarium’s (CC BY-SA 4.0), as published. The Moon is drawn with its lit fraction, its lit side facing along the great circle toward the Sun. “Point with my phone” takes its direction from the phone’s compass, which points to magnetic north; the map turns that to true north for its place with NOAA’s World Magnetic Model 2025 (valid 2025 to 2030, and checked against NOAA’s 100 published test values). What it cannot correct is the phone’s own compass error near metal or electronics. Nebulae, clusters and galaxies are Messier’s 110 and four more the guides name, with positions, types, sizes, magnitudes where it has one, and names from the SIMBAD database (CDS, Strasbourg). Meteor shower radiants are drawn while the IMO’s calendar says each shower is active, following the radiant’s drift from its table of positions every five days. “Air’s bending”, off by default, lifts positions to where they appear: Bennett’s refraction formula for 10 °C and 1010 hPa, as Skyfield computes it (it matches Skyfield to a thousandth of an arcsecond), about half a degree at the horizon; everywhere else on the site altitudes stay geometric. Fourteen observing windows agree to within seconds. Those reference values are part of the site’s build checks, which is also why the finder is limited to 2000–2050.

A suggestion needs the Sun at least 12° below the horizon and the star at least 20° up, or, for a constellation, the middle of its key pattern 20° up with every anchor star above the horizon. These are this site’s rule of thumb for a useful suggestion, not physical limits. A night runs from local noon to local noon. Directions are rounded to 5° of azimuth and whole degrees of altitude, and nothing accounts for hills, buildings, cloud, haze, light pollution or your eyes. City coordinates and time zones come from the GeoNames gazetteer (CC BY 4.0), about 8,000 places.

The planets

Where the planets, the Sun and the Moon appear is worked out in your browser by this site’s own engine. The planets come from VSOP87, the planetary theory of Bretagnon and Francou (1988), trimmed for 1900–2050 until it stays within a tenth to six tenths of an arcsecond of the full theory; Pluto from a fit to NASA JPL’s DE421 ephemeris; the Moon from the same Meeus series as the rest of the site. The apparent position adds the time light takes to reach us, aberration and nutation, and uses Terrestrial Time, with the difference from clock time taken from the International Earth Rotation Service’s measurements and a published model beyond them.

It is checked against Skyfield with DE421, a different calculation of the same thing, at 160 moments from 1900 to 2050: the Sun agrees within half an arcsecond, Mercury to Saturn within one, Uranus and Neptune within three, and the Moon within eleven (about a hundred-and-seventieth of its own width). Brightness uses the Astronomical Almanac’s formulas and is within a few tenths of a magnitude of Skyfield’s, which is enough to say whether you need binoculars. Outside 1900–2050 the site declines to answer rather than extrapolate.

Friends’ Skies

Friends’ Skies first lists what is happening at all: the Moon’s phases and its first and last thin crescents; the Moon passing within seven degrees of Mercury, Venus, Mars, Jupiter or Saturn, and two of those within three degrees of each other, found as closest approaches in the positions above; Mars, Jupiter and Saturn (and, for binoculars and telescopes, Uranus and Neptune) opposite the Sun; Mercury and Venus at their greatest distance from it; and the eclipses and meteor showers described below, from NASA and the International Meteor Organization.

Then it asks, for each saved place separately, whether and when each one is visible there, with rules that suit the event rather than one height for everything. A pairing needs both objects at least 5° up (3° for Mercury) with the Sun at least 4° to 6° below the horizon; a thin crescent is judged at the moment the Sun is 6° down; a planet at opposition by how high it gets while the Sun is at least 9° down; a meteor shower by the height of its radiant while the Sun is at least 12° down and by how much of the Moon is lit; a lunar eclipse by the Moon’s height at its middle; a solar eclipse by whether the Moon’s shadow reaches the place at all, with the same shadow geometry as the eclipse pages, and the Sun’s height then. Heights include the Moon’s parallax, which is up to a degree. The result is labelled Excellent, Good or Challenging, or not offered at all, and ranked as “worth telling someone”, “worth a look” or quieter by a fixed rule for each kind of event, so a quiet week is shown as quiet. Times are the place’s own clock and are given to five minutes. Weather is not part of any of it.

Birth charts

The birth chart is built in three layers that are kept apart in the code and on the page. The astronomy: apparent geocentric positions from the engine described above, for the instant your local birth time corresponds to, using the time-zone history of the birthplace. The mapping: tropical signs (30° slices from the March equinox), houses and angles, aspects and their orbs, each by a stated rule. The interpretation: text written for this site in the reflective voice of modern Western astrology, which is tradition, not a finding.

The Ascendant, Midheaven and the house cusps (Placidus, Whole Sign or Equal) have been checked against the Swiss Ephemeris, the reference most astrology software is measured against, run separately on our own computer: at 71 test charts from 1900 to 2050 and from 60° south to 78° north, every cusp and angle agrees within three arcseconds, the planets within about one and the Moon within nine. Placidus cannot be drawn near the poles, where some degrees never rise or set; the page says so and lets you choose Whole Sign or Equal instead, as the Swiss Ephemeris also refuses. Aspects use orbs of 8° for conjunctions and oppositions, 7° for trines and squares and 5° for sextiles, 2° more with the Sun or Moon and 3° fewer to the Ascendant or Midheaven. Degrees are truncated, as ephemerides print them.

With an approximate time the chart is also calculated at both ends of the range, and anything that changes is marked beside it. With no time there is no Ascendant, no houses and no angles; the Moon's range across the day is shown, and aspects that come or go during the day are left out. Dates before about 1970 depend on reconstructed time-zone records, so a manual UTC offset is offered. Chiron, asteroids, Black Moon Lilith and the lots are defined in the astrology dictionary but not placed in the chart.

Daily astrology

The daily readings use the same engine and the same three layers. A transit is found by stepping each planet through time more finely than any transit could slip between two steps, then narrowing every crossing down to the second; the results agree with the Swiss Ephemeris and with NASA JPL’s DE421 to within a few arcseconds of each planet’s motion. The orbs, the ranking that picks a day’s reading, and the rules for uncertain birth times are set out in full on how your daily sky is worked out.

Astronomy words

The astronomy dictionary and the Astronomy Word of the Day are written for this site in its own words. They are single words that astronomers really use and that are seldom explained, not phrases or abbreviations. Every word cites sources chosen for it, mostly from NASA, ESA, ESO and other scientific institutions, and every link is checked to open. Each pronunciation is a respelling and a broad General American IPA for the whole word; other accents say some words differently. Each date’s word comes from a schedule fixed a year ahead, so a date’s word never changes; the words take turns, and each one comes round again only after all the others have had a day. Every word has a “Report a problem” link: if a definition, a pronunciation or a source looks wrong, it is the fastest way to get it fixed.

Your location, and how precise it needs to be

Rise and set times need a position. A location you set is rounded to two decimal places — a little over a kilometre — because that is already finer than the difference this calculation can honestly resolve, and because a rounder number is a smaller thing to be keeping. Town coordinates come from the GeoNames gazetteer (CC BY 4.0); the list of about 21,500 places is stored on this site and searched on your device. If you set no location, the site hides rise and set times rather than guessing.

What the times do not account for

Standard atmospheric refraction is included. Your actual horizon is not. A hill, a building, a treeline or an unusual air mass will move a real moonrise by minutes, and at high latitudes by a great deal more. Treat the times as “when it clears a flat sea horizon”, which is what they are.

Eclipses

Dates, times and durations are NASA’s, from the Five Millennium Canons of Solar and Lunar Eclipses. Everything else on those pages is this site’s own: the shadow track is computed by intersecting the Sun–Moon shadow cone with an Earth modelled at its real flattened shape, and drawn as a globe when the site is built. The computed length of totality agrees with NASA’s published figure to within one percent on all eight central solar eclipses in the range, which is the check that makes the track worth publishing. Refraction and terrain are not modelled. For whether a track crosses your street, use NASA’s own maps (opens in a new tab), not this site.

Meteor showers

Peak dates and rates are the International Meteor Organization’s. What is computed here is the part that changes every year: where the Moon will be on the peak night and how much of the shower it will wash out. Published rates are zenithal hourly rates — what a perfect observer would see under a perfect sky with the radiant overhead. Nobody has ever seen that many. Expect a fraction of it. How high the radiant gets is computed here too, from its position in the IMO calendar and the Sun’s, for the peak night, counting only the hours when the Sun is at least 12° below the horizon. That last part matters: the Eta Aquariid radiant is highest after sunrise, so from 40°N it reaches about 23° while it is dark, not the 49° a textbook formula gives. The share of meteors you would see at a given height is the sine of that height, the same correction the IMO uses to turn a real count into a zenithal hourly rate.

The Moon you are looking at

Rendered from Lunar Reconnaissance Orbiter data: colour from the LROC wide-angle mosaic (NASA/GSFC/Arizona State University), relief from the LOLA global elevation model (NASA/LOLA Science Team). The frames are pre-rendered when the site is built, so displaying the Moon fetches nothing from anyone. Sixteen of the same frames make the Moon that goes through its phases while a slow job (a birth chart, several friends’ skies) is still working. It appears only if the work takes longer than a quarter of a second, goes the moment the work is done, and holds still if your device asks for less motion. It is a rendering, not tonight’s Moon.

The stars behind the page

Real ones, down to magnitude 6.5, from the Bright Star Catalogue (Hoffleit & Warren 1991) via VizieR/CDS, plotted around your zenith for the current moment. The coastlines on the eclipse globes are Natural Earth 1:110m, which is public domain and is credited anyway.

Folklore, history and astrology are marked as such

The folklore, the moon names, the historical events and the astrology material are kept visibly separate from the computed astronomy, and are presented as what people have said and done rather than as measured effects. The astrology and myths pages go one step further and label every interpretation as modern and symbolic, and every reflection prompt as this site's own. The phase they start from is the same computed phase as everywhere else; the mode never changes a number. Where an attribution is commonly repeated and thin — most of the “traditional” full moon names are — the page says so instead of passing it along. Historical claims are sourced to primary records and reference works where they exist.

How often it is checked

Every fact card carries the date it was last verified, and the corpus is swept in full periodically rather than on a fixed schedule; the tonight page’s sources sheet shows the date of the most recent sweep. The eclipse and shower tables are static data with a fixed range and do not go stale within it. If you find something wrong, tell us — that is the fastest correction mechanism this site has.

Where this site should not be your source

Anything you are travelling for, timing an observation against, or making a decision on. Use NASA for eclipse circumstances, a national weather service for sky conditions, and an ephemeris service if you need positions to arcsecond precision. This site is built to be right enough to plan an evening around and to be honest about the difference.