How Greenwich time — and therefore longitude — was obtained by measuring the angle between the Moon and the Sun
1. The problem: a clock nobody had
Longitude is a difference in time. If I know what time it is at Greenwich and, by observing the sky, I work out local time, the difference between the two gives me my longitude: one hour is fifteen degrees. Throughout the eighteenth century the difficulty was not measuring the sky — sextants and almanacs existed — but the clock: no mechanism survived months of sea, damp and rolling without losing minutes, and one minute of error is fifteen minutes of arc of longitude.
Two solutions emerged, and ran side by side for fifty years. The first was mechanical: John Harrison’s chronometers, followed by the far cheaper instruments of Arnold and Earnshaw. The second was astronomical and asked nothing of any mechanism: use the Moon as the hand of a clock built into the sky.
2. The idea: the Moon as a clock hand
The Moon moves among the stars by about half a degree an hour — roughly its own apparent diameter. The angle between the Moon and the Sun, or between the Moon and a star near the ecliptic, therefore changes continuously and predictably. It is a function of time, the same for every observer on Earth, because it is computed from the centre of the Earth.
Hence the method. From 1767 Nevil Maskelyne’s Nautical Almanac tabulated, every three hours, the geocentric distance between the Moon and a set of chosen bodies. The navigator measured the distance between the two bodies with a sextant, reduced it to the centre of the Earth, and looked in the table for the time at which that distance occurred: that time was Greenwich time. With local time from an altitude of the Sun, longitude followed by subtraction.
| The price. The Moon moves 30′ an hour, that is 1′ every two minutes of time. An error of 1′ in the measured distance becomes an error of 2 minutes in Greenwich time, and so about half a degree of longitude: thirty miles at the equator. The method therefore demands measurements good to a tenth of a minute of arc, on an angle of seventy or ninety degrees, taken from a moving ship. |
This is why lunar distances pushed instruments to their limits, and it was on this ground that Jean-Charles de Borda left his mark twice: with an instrument and with a method of calculation.
3. Borda: the instrument and the method
Jean-Charles de Borda (1733–1799) was a naval officer, mathematician and member of the Académie des sciences. He served in the American War of Independence, was captured by the English in 1782 and appointed Inspector of Naval Shipbuilding in 1784. He first used Tobias Mayer’s reflecting circle in 1771, during a cruise aboard the frigate La Flore. [B]
The reflecting circle has one advantage over the sextant: the graduated circle is complete, so the same measurement can be repeated many times, accumulating round the limb, without returning the index to zero. The random error of each single reading is divided by the number of repetitions. Between 1775 and 1777 Borda refined the design, which from 1783 was built by Étienne Lenoir: by repetition, the « Borda circle » reached resolutions of the order of one minute of arc. [B]
The instrument alone was not enough. Reducing a lunar distance to the centre of the Earth required a long computation, and every step done by hand was a chance to go wrong. Borda published his reduction method in 1778 and returned to it in « Description et usage du cercle de réflexion, avec différentes méthodes pour calculer les observations nautiques ». Its success is explained in one line: the method is entirely logarithmic. There is not one sum or difference of products; one adds and subtracts logarithms of cosines read from the tables, halves the total, and enters the tables backwards only once. For an officer working at the chart table in a seaway, that was the difference between a possible computation and an impossible one. [B] [A]
A note on dates. The year 1770 often associated with Borda belongs to the « Borda count » voting method, not to lunar distances. For the reduction method the sources give 1778; the treatise on the reflecting circle came later and ran to several editions down to 1816. [B]
4. Why the measured distance is not the tabulated one

Fig. 1 — The measured distance D joins the apparent positions; the tabulated distance D′ joins the true ones. The displacements are exaggerated for clarity.
The almanac gives the distance as seen from the centre of the Earth, between the centres of the two bodies, with no atmosphere. The sextant measures something quite different:
- refraction raises both bodies above their true positions, and the more so the lower they are;
- parallax lowers the Moon, and lowers it a great deal: observed from the surface instead of from the centre of the Earth, the Moon can appear almost a degree lower;
- the sextant measures between limbs, not between centres;
- a Moon near the horizon looks slightly larger than the almanac states, because the observer is one Earth radius closer to it (augmentation of the semi-diameter).
Correcting altitudes is everyday work. The delicate point is the distance: parallax and refraction move the two bodies along their own vertical circles, that is in different directions, and so distort the angle between them. Going from the apparent distance D to the true distance D′ is called clearing the distance, and it is the heart of the problem.
4.1 The geometric key
One thing does not change: the angle Z at the zenith, the angle between the two vertical circles through the bodies. Refraction and parallax move each body along its own vertical circle, not sideways. The spherical triangle Z–Moon–Sun therefore has the same angle at Z before and after correction, and of that triangle we know the three apparent sides and two true sides: it is enough to eliminate Z between the two forms of the cosine rule.
cos D = sin ha(M) · sin ha(S) + cos ha(M) · cos ha(S) · cos Z
cos D′ = sin H(M) · sin H(S) + cos H(M) · cos H(S) · cos Z
Here ha are the apparent altitudes (as measured, corrected only for index error, dip and semi-diameter) and H the true altitudes (also corrected for refraction and parallax). Taking cos Z from the first and substituting it in the second gives the modern solution, fit for any calculator [C]:
cos D′ = [ cos D + cos ( ha M + ha S ) ] · cos H(M) · cos H(S) / [ cos ha(M) · cos ha(S) ] − cos ( H M + H S )
5. Borda’s formula
The formula above contains a sum and a difference: logarithms cannot handle those. Borda rewrites it with the half-angle identities so that everything becomes a product or a quotient, and introduces a single auxiliary angle θ [B] [C]:
s = ½ ( D + ha M + ha S )
cos θ = √[ cos s · cos ( s − D ) · cos H(M) · cos H(S) / ( cos ha M · cos ha S ) ] / cos ½ ( H M + H S )
sin ½ D′ = cos ½ ( H M + H S ) · sin θ
In practice one adds six logarithms of cosines (four positive, two negative), halves the result — that is the square root — subtracts a seventh logarithm, and has log cos θ. A second line of two logarithms gives sin ½D′. No multiplication, no root to extract by hand: tables only.

Fig. 2 — The chain of corrections: altitudes from sextant to true, distance from reading to cleared.
6. Worked example
A simulation built for this article: the observations are computed from the ephemerides for a known position, then rounded to a tenth of a minute as a real observer would read them. The error that appears at the end is therefore the error of the reduction method alone. [C]
6.1 The data
| Item | Value |
| Date | 16 October 2026 |
| Dead reckoning position | 36° 00.0′ N — 010° 00.0′ W (longitude is what we want to check) |
| Chronometer | unreliable: all that is known is that it is « about 15h » Greenwich |
| Height of eye ; index correction | 3 m ; −1.2′ |
| Measured distance, near limbs Moon–Sun | 67° 43.7′ |
| Altitude of Moon’s lower limb | 19° 41.9′ |
| Altitude of Sun’s lower limb | 30° 43.0′ |
| From the almanac | Moon HP 54.2′ ; Moon SD 14.8′ ; Sun SD 16.0′ |

Fig. 3 — At the sextant the limbs are brought into contact: the two semi-diameters are added to the reading.
6.2 From readings to apparent altitudes and apparent distance
| Moon | Sun | Distance | |
| Sextant reading | 19° 41.9′ | 30° 43.0′ | 67° 43.7′ |
| Index correction | −1.2′ | −1.2′ | −1.2′ |
| Dip (3 m) | −3.0′ | −3.0′ | does not apply |
| Semi-diameter (Moon: 14.8′ + 0.1′ augmentation) | +14.9′ | +16.0′ | +14.9′ and +16.0′ |
| Result | ha = 19° 52.6′ | ha = 30° 54.8′ | D = 68° 13.4′ |
Dip does not enter the distance: both bodies are in the sky and the horizon plays no part. The index correction enters only once, because there is only one reading.
6.3 From apparent to true altitudes
| Moon | Sun | |
| Apparent altitude ha | 19° 52.6′ | 30° 54.8′ |
| Refraction | −2.7′ | −1.7′ |
| Parallax (HP · cos ha ; Sun 0.1′) | +51.0′ | +0.1′ |
| True altitude H | 20° 40.9′ | 30° 53.2′ |
The asymmetry of the problem shows at once: the Sun comes down by a minute and a half, the Moon goes up by almost a degree. It is this difference that distorts the measured angle.
6.4 The reduction by Borda’s formula
ha M + ha S = 50° 47.4′ H M + H S = 51° 34.1′ ½ ( H M + H S ) = 25° 47.0′
s = ½ ( D + ha M + ha S ) = 59° 30.4′ s − D = − 8° 43.0′
| Logarithm (five figures as in the tables, characteristic + 10) | Value |
| log cos s = log cos 59° 30.4′ | 9.70538 |
| log cos ( s − D ) = log cos 8° 43.0′ | 9.99495 |
| log cos H(Moon) = log cos 20° 40.9′ | 9.97107 |
| log cos H(Sun) = log cos 30° 53.2′ | 9.93358 |
| − log cos ha(Moon) = log cos 19° 52.6′ | − 9.97332 |
| − log cos ha(Sun) = log cos 30° 54.8′ | − 9.93346 |
| sum | 19.69820 |
| half the sum (this is the square root) | 9.84910 |
| − log cos ½ ( H M + H S ) = log cos 25° 47.0′ | − 9.95445 |
| log cos θ | 9.89465 → θ = 38° 19.0′ |
| log sin θ + log cos ½ ( H M + H S ) | 9.79239 + 9.95445 = 9.74684 |
| sin ½ D′ | 0.55826 → ½ D′ = 33° 56.2′ |
The sign of s − D does not matter: the cosine of a negative angle equals the cosine of its absolute value.
| Cleared distance: D′ = 67° 52.3′ With a calculator, without rounding to the five decimals of the tables, the same formula gives 67° 52.4′. The exact geocentric distance at that instant, from the ephemerides, is 67° 52.4′: the error of the reduction method is 0.04 minutes of arc, that of five-figure logarithmic tables about 0.1 minutes. [C] [V] |
6.5 From Greenwich time to longitude
The modern almanac no longer tabulates lunar distances, but they are computed in one line from the data it does publish, with the same cosine rule used for every spherical triangle [C]:
cos D = sin Dec(M) · sin Dec(S) + cos Dec(M) · cos Dec(S) · cos [ GHA(M) − GHA(S) ]
| UT on 16/10/2026 | GHA Moon | Dec Moon | GHA Sun | Dec Sun | Geocentric distance |
| 12h | 295° 33.1′ | S 27° 47.7′ | 003° 36.5′ | S 8° 59.7′ | 66° 27.5′ |
| 15h | 339° 00.3′ | S 27° 43.3′ | 048° 36.9′ | S 9° 02.4′ | 67° 48.6′ |
| 18h | 022° 27.7′ | S 27° 37.7′ | 093° 37.3′ | S 9° 05.2′ | 69° 09.6′ |

Fig. 4 — The distance grows by 27′ an hour: enter with the cleared distance, come out with Greenwich time.
Between 15h and 18h the distance goes from 67° 48.6′ to 69° 09.6′, that is 81.0′ in 180 minutes. The cleared distance exceeds the 15h value by 3.8′. Therefore [C]:
3.8′ / 81.0′ × 180 min = 8.4 min GMT = 15h 08m 25s
The true time of the simulation was 15h 08m 20s: five seconds of error. Local time remains to be found. The Sun’s true altitude is 30° 53.2′; with declination S 9° 02.4′ and latitude 36° 00.0′ N, the hour-angle formula gives [C]:
cos LHA = ( sin H − sin φ · sin Dec ) / ( cos φ · cos Dec ) LHA = 40° 42.0′ (afternoon)
λ = LHA − GHA(Sun) = 40° 42.0′ − 50° 43.2′ = − 10° 01.2′ that is 010° 01.2′ W
| Result: 36° 00′ N — 010° 01.2′ W, against the true 010° 00.0′ W: one mile of error. Almost all of it comes from the five seconds of error in Greenwich time, which are worth 1.2′ of longitude. [C] [V] |
7. What it is really worth at sea
The one mile of the example is a credit to the method of calculation, not a promise of accuracy. At sea the dominant error is not arithmetic but observational: measuring an angle of seventy degrees between two bodies to a tenth of a minute, from a deck that moves, is hard. Half a minute of error in the distance — already a good measurement — means about one minute in Greenwich time, that is a quarter of a degree of longitude, fifteen miles at the equator. [B]
The conditions for a good lunar are few and strict: [A]
- distance between the two bodies between about 40° and 120°: narrower, the geometry is weak; wider, the measurement is awkward;
- altitudes of both bodies above 15°–20°, to keep refraction and its uncertainty small;
- the two altitudes and the distance observed as nearly simultaneously as possible, or reduced to the same instant;
- repetition: three or five distances averaged are worth far more than a single one.
This explains the end of the method. As soon as chronometers became reliable and cheap, nobody missed lunar distances: the British Nautical Almanac printed the tables until 1906, and then stopped. [B]
8. Why it is still worth knowing
Because it is the only way, now as then, to get Greenwich time out of the sky with a sextant and an almanac and no clock at all. A chronometer that stops, a silent radio and a dead GPS leave no other road. And because the procedure teaches, better than any exercise, what parallax and refraction really are: not two corrections to be looked up, but two distortions that shift the measured sky away from the computed one.
Anyone trying it today has an advantage over Borda’s contemporaries: the tabulated distances can be computed in one line from the current Nautical Almanac with the cosine rule of section 6.5; and to clear the distance one may choose between Borda’s formula, still the most elegant, and its modern calculator form. The sextant, on the other hand, has not changed.
Sources and notes
| Source | What was taken from it |
| Jean-Charles de Borda — en.wikipedia.org/wiki/Jean-Charles_de_Borda | Life dates, naval career, publication of the lunar-distance reduction method (1778), trigonometric tables published in 1801. [B] |
| Cercle de réflexion — fr.wikipedia.org/wiki/Cercle_de_réflexion_(instrument) | Tobias Mayer’s circle (1752, described 1767), Borda’s use of it aboard La Flore in 1771, refinement 1775–1777, Lenoir’s construction from 1783, resolution by repetition. [B] |
| Lunar distance (navigation) — en.wikipedia.org/wiki/Lunar_distance_(navigation) | Tables published from 1767 to 1906, the Moon’s motion of 0.5° per hour, the error budget (half a minute of distance ≈ one minute of time ≈ 15 miles). [B] |
| HM Nautical Almanac Office — astro.ukho.gov.uk/nao/history/nao_1767.html | The first Nautical Almanac (for 1767), its stated purpose of lunar distances, Maskelyne’s Tables Requisite. [B] |
| Borda, « Description et usage du cercle de réflexion » — gallica.bnf.fr | Borda’s treatise on the circle and on methods of computing nautical observations, available online. [B] |
| The author’s own work | Derivation of the reduction formula, numerical simulation of the example, computation of the ephemerides (IAU/ERFA algorithms) and check of the residual error. [C] |
Key: [B] bibliographic source · [C] the author’s computation or derivation, numerically verified · [V] cross-check · [A] the author’s assessment.
0 commenti