Moon Phase Calculator

lunar phase, illumination, and orbital geometry by date

Loading interactive simulation...

Why the full Moon looks full for three nights 🖖

Illumination here is a cosine of the Moon's age, not a straight line: the tool computes I = (1 βˆ’ cos 2πφ)/2, where Ο† is the fraction of the 29.53-day synodic month elapsed. A cosine is flat at its turning points and steepest between them, so brightness barely moves near full and sprints through the quarters. Step the date back one day from the next full moon and illumination falls only from 100% to 98.9%, a change no eye will catch. Do the same from first quarter and it swings from 50% to 60.6%, nine times as much. That is why a full Moon looks full for about three nights running while a crescent visibly thickens from one night to the next.

Reading the Moon's age off the sky 🖖

This tool's 'moon age' is simply the number of days since the last new moon, counting up to about 29.5. That single number fixes the phase: a young Moon is a thin crescent, day 7 is first quarter, day ~15 is full. In the northern hemisphere a growing (waxing) Moon is lit on its right side; in the southern hemisphere it is flipped, which is why the tool offers a hemisphere switch.

We see 59% of a tidally locked Moon 🖖

The Moon keeps one face toward us because its spin and orbit are locked to the same 27.3-day period. You would expect exactly half its surface to be forever hidden β€” yet over a month the Moon appears to nod and rock slightly, an effect called libration, because its orbit is elliptical and tilted. These wobbles let us peek around the edges, so about 59% of the lunar surface is visible from Earth over time, not 50%.

Problem solved in full

  1. Deriving the synodic month of 29.5306 d from the sidereal month 6 steps

    Seven of the table's ten rows put a formula in the formula column. The Synodic month row puts a number: S = 29.5306 d, asserted. Derive it instead β€” from the Sidereal month row directly below it, plus one given the tool never mentions.

    1. A month is a race between two angular speeds, so start by converting one. The sidereal month is the Moon's full lap against the fixed stars, and that is a clean 360Β° in the printed 27.3217 d.

    2. Phase, though, is not measured against the stars β€” it is measured against the Sun, and the Sun's direction is not fixed, because Earth is orbiting it. The Sun therefore drifts one full turn against those same stars per sidereal year, 365.2564 d. This is the one number that does not come from the page; you have to bring it.

    3. Now run one sidereal month and see what is missing. The Moon is back beside its starting star, but the Sun has slid on by ΞΈ0, and the Moon needs a further t1 to close that gap and be lit the same way again.

    4. The Sun does not wait during the catch-up. In those 2.0437 d it moves again, by the fraction r of the gap it had already opened, and again during the fix for that β€” a geometric series, which is why the true delay is t1 divided by 1 βˆ’ r rather than t1. Because r is only 0.075, the whole infinite chase costs just 8% more than the first estimate.

    5. Add the catch-up to the sidereal month β€” or skip the series entirely, since phase advances at the difference of the two rates and 360Β° over that difference gives the same answer. Both routes land on 29.5306 d, which is precisely what the Synodic month row asserts, and 29.531 d is what its value column prints.

    6. Then check the arithmetic a second way, using nothing from outside the table at all: subtract the two printed values.

    Answer

    S = 29.5306 d, and it is not a fact about the Moon β€” it is 27.3217 d of orbit plus 2.2089 d of chasing a Sun that moved. The gap between the table's two "month" rows is that chase, which is why they differ by 2.209 d and not by something arbitrary. Push the result somewhere the tool cannot go: twelve synodic months are 12 Γ— 29.5306 = 354.367 d, which falls 10.875 d short of the 365.2422 d tropical year β€” the year the seasons actually follow, slightly shorter than the sidereal year of step 2 because the equinox itself creeps backwards. So a purely lunar calendar's months walk backwards through the seasons and return to the same season only after 365.2422 / 10.875 = 33.6 lunar years — which is 32.6 solar ones, and the near-miss between those two counts is exactly the drift being measured. Either way it is how Ramadan crosses summer and winter within one lifetime. But go to 235 months instead of 12 and the mismatch almost disappears: 235 Γ— 29.5306 = 6939.69 d against 19 tropical years of 6939.60 d, agreeing to 0.09 d β€” about two hours. Moon phase on a given calendar date therefore repeats every 19 years, which is the Metonic cycle, and the reason Easter can be computed centuries in advance rather than watched for.

References (3)