Light Left

Every star is a date

Each star on these two plates carries the year its light left it, worked out from the star's measured parallax. Every year is a range, because every parallax has an error. Where the error is too large to hold a year, the star stays blank.

Label each star with
4 years ago404004,000
Open ring: parallax too uncertain to date

Each year is when the star's light left it, as a range.

North The pole at the centre; the rim 20° south of the equator
South The pole at the centre; the rim 20° north of the equator

Point at a star, or tap it, for its record.

Reading the Hipparcos catalogue.

The Birth Star

The light that left the day you were born

Enter a birth date to find the star whose light left that day. The date stays on this device. Nothing is sent anywhere.

The Radio Bubble

How far Earth's own broadcasts have travelled tonight

Radio has been leaving Earth since the evening of 2 November 1920.

    The Method

    Parallax, its error, and why the range exists

    Parallax in One Picture

    Earth at two ends of its orbit, six months apart, sees a near star against different parts of the far background; the angle at the star is twice the parallax. January July Sun 2p a near star the far background

    Earth swings two astronomical units, about 299 million kilometres, across its orbit every six months. Seen from the two ends of that swing, a nearby star shifts against the far background by twice its parallax. A parallax of one arcsecond puts a star at one parsec, 3.26156 light-years, and the distance in light-years is the number of years the light has been travelling. Vega's parallax is 130.23 milliarcseconds, so Vega is 25.04 light-years out, and its light set off 25.04 years before it reaches you.

    Why the Range

    Hipparcos measured each parallax with an error, and the catalogue prints both: Vega's is 130.23 ± 0.36 milliarcseconds. A parallax one error larger puts the star nearer; one error smaller, farther. This page dates both edges, so every year it prints is the span between them. Near stars are known to within weeks; past a few hundred light-years the span runs to centuries.

    When the error is more than a fifth of the parallax the far edge runs away, and at an error equal to the parallax it reaches infinity. Those stars get no year. Alnilam, the middle star of Orion's belt, has a parallax of 1.65 ± 0.45 milliarcseconds: somewhere between 1,553 and 2,718 light-years. The chart draws it as an open ring.

    Why Hipparcos

    ESA's Hipparcos satellite measured the parallaxes of 118,218 stars between 1989 and 1993, and Floor van Leeuwen's 2007 reduction of its data is the catalogue here: every star brighter than magnitude 7, plus every star within about 163 light-years. Gaia, its successor, measures faint stars far better but saturates on the brightest: a fifth of stars brighter than magnitude 3 have no entry in its early third data release, Polaris among them (Fabricius et al. 2021). For the stars you can see, Hipparcos is the better ruler.

    The Constants

    1 parsec is 648,000/π astronomical units, 3.0856775814913673 × 1016 metres (IAU 2012). 1 light-year is the distance light covers in a Julian year of 365.25 days, 9.4607304725808 × 1015 metres. So 1 parsec is 3.2615637771674337 light-years. Dates count back from tonight in Julian years, on the Gregorian calendar run backwards past 1582; years before 1 CE print as BCE.

    The Receipts

    Twelve distances from this page's copy of the catalogue, against the same stars' parallaxes read separately from SIMBAD, the CDS database, each citing van Leeuwen (2007).

    StarThis page, lySIMBAD, lyDifference

    Constellations follow N. G. Roman's 1987 table of the IAU boundaries; her eight published test positions are checked when the page loads.

    What It Cannot Know

    • The range is one standard error. The true distance falls outside it about one time in three.
    • Hipparcos is complete only to about magnitude 7.3. Faint red dwarfs within a hundred light-years are missing, so every count of stars inside the radio bubble is a floor.
    • Positions are the catalogue's, for 1991.25. The stars have moved a little since, invisibly at this scale.
    • The radio bubble is a sphere of time, not of signal: it says how far the first broadcasts have travelled, not whether anyone there could hear them.
    • Fessenden's Christmas Eve 1906 programme is known only from his own account, written in 1932, and historians dispute it.