Several examples of the value of eclipses in chronology are mentioned above in passing. No one system of dating has been continuously in use since ancient times, although some, such as the Olympiads, persisted for many centuries. Dates were frequently expressed in terms of a king’s reign; years were also named after officials of whom lists have been preserved (for instance, the Assyrian Chronicle mentioned above). In such cases, it is important to be able to equate certain specific years thus defined with years before the Christian era (bce). This correspondence can be made whenever the date of an eclipse is given in an ancient record. In this regard, eclipses have distinct advantages over other celestial phenomena such as comets: in addition to being frequently recorded in history, their dates of occurrence can be calculated exactly.

Chinese chronology can be confirmed accurately by eclipses from the 8th century bce (during the Zhou dynasty) onward. The Chunqiu chronicle, mentioned above, notes the occurrence of 36 solar eclipses between 722 and 481 bce—the earliest surviving series of solar eclipse observations from any part of the world. The records give the date of each event in the following form: year of the ruler, lunar month, and day of the 60-day cycle. As many as 32 of the eclipses cited in the Chunqiu can be identified by modern calculations. Errors in the recorded lunar month (typically amounting to no more than a single month) are fairly common, but both the year and the recorded day of the sexagenary cycle are invariably correct.

The chronology of Ptolemy’s canon list of kings—which gives the Babylonian series from 747 to 539 bce, the Persian series from 538 to 324 bce, the Alexandrian series from 323 to 30 bce, and the Roman series from 30 bce onward—is confirmed by eclipses. The eclipse of 763 bce, recorded in the Assyrian Chronicle, makes it possible to carry the chronology back with certainty through the period covered by that eponym canon to 910 bce. Identifiable eclipses that were recorded under named Roman consuls extend back to 217 bce. The lunar eclipse seen at Pydna in Macedonia on June 21, 168 bce, and the solar eclipse observed at Rome on March 14, 190 bce, can be used to determine months in the Roman calendar in the natural year. Furthermore, eclipses occasionally help to fix the precise dates of a series of events, such as those associated with the Athenian disaster at Syracuse in 413 bce.

The late Babylonian astronomical texts occasionally mention major historical events, as, for example, the dates when Xerxes and Alexander the Great died. To illustrate the potential of this material for chronological purposes, the date of the death of Xerxes may be accurately fixed by reference to eclipses. On a tablet that lists lunar eclipses at 18-year intervals occurs the following brief announcement between two eclipse records: “Month V, day 14 [?], Xerxes was murdered by his son.” Unfortunately, the cuneiform sign for the day of the month is damaged, and a viable reading could be anything from 14 to 18. The year is missing, but it can be deduced from the 18-year sequence as 465 bce. This identification is confirmed by calculating the dates of the two eclipses stated to have occurred in the same year that Xerxes died. The first of these happened when the Moon was in the constellation of Sagittarius, while the second took place on the 14th day of the eighth lunar month. For many years both before and after 465 bce, no such combination of eclipses can be found; it occurs only in 465 bce itself. The dates deduced for the two eclipses are June 5 and November 30 of that year. Mention of an intercalary sixth month on the same tablet enables the date of the death of Xerxes to be fixed as some time between August 4 and 8 in 465 bce.

Uses of eclipses for astronomical purposes

Ancient and medieval observations of eclipses are of the highest value for investigating long-term variations in the length of the day. Early investigators such as the English astronomer Edmond Halley deduced from eclipse observations that the Moon’s motion was subject to an acceleration. However, not until 1939 was it conclusively demonstrated (by the British astronomer Harold Spencer Jones) that only part of this acceleration was real. The remainder was apparent and was a consequence of the practice of measuring time relative to a nonuniform unit, namely, the rotation of Earth. Time determined in this way is termed Universal Time. For astronomical purposes, it is preferable to utilize an invariant time frame such as Terrestrial Time (the modern successor to Ephemeris Time)—defined by the motion of the Sun, Moon, and planets.

Lunar and solar tidal friction, occurring especially in the seas and oceans of Earth, is now known to be responsible for a gradual decrease in the terrestrial rate of rotation. Apart from slowing down Earth’s rotation, lunar tides produce a reciprocal effect on the Moon’s motion, causing a gradual increase in the mean distance of the Moon from Earth (at about 3.8 cm [1.5 inches] per year) and a consequent real retardation of its motion. Hence, the length of the month is slowly increasing (at about 0.04 second in a century). (See also Moon: Principal characteristics of the Earth-Moon system.)

These changes in the Moon’s orbit can now be accurately fixed by lunar laser ranging, and it seems likely that they have proceeded at an essentially constant rate for many centuries. The history of Earth’s rotation, however, is complicated by effects of nontidal origin, and in order to obtain maximum information it is necessary to utilize both modern and ancient observations. Telescopic observations reveal fluctuations in the length of the day on timescales of several decades, and these fluctuations are mainly attributed to interactions between the fluid core of Earth and the surrounding solid mantle. Ancient and medieval observations also suggest the presence of longer-term variations, which could be produced by alterations in the moment of inertia of Earth resulting from both the ongoing rise of land that was glaciated during the Pleistocene Ice Age (which ended around 11,700 years ago) and the sea-level changes associated with the freezing and melting of polar ice.

Records of large solar eclipses preserved in literary and historical works have made an important contribution to the study of past variations in Earth’s rate of rotation. In recent years, major advances have also come from the analysis of ancient and medieval timings of lunar and solar eclipses by Babylonian and Chinese and Arab astronomers. Although many Babylonian texts are fragmentary, about 120 usable timings of eclipse contacts are accessible (including measurements at different phases of the same eclipse). These observations date primarily from between about 700 and 50 bce. By comparison, only a handful of similar Greek measurements are preserved, and these are less precise. Approximately 80 eclipse timings by Chinese astronomers are preserved in Chinese history. These are from two main periods: between 400 and 600 ce and later from 1000 to 1300. In addition, almost 50 measurements of eclipse times by medieval Arab astronomers are extant; these date from between about 800 and 1000 ce and are mainly contained in the Hakemite Tables compiled by Ibn Yūnus about 1005. Unfortunately, there are very few timings between 50 bce and 400 ce and again from 600 to 800.

Tidal computations indicate a steady increase in the length of the mean solar day by about 1/40second every millennium. Nontidal causes produce smaller effects, generally in opposition to the main trend. Although the rate of change in the length of the day is minute, the loss of energy by Earth is huge. In measuring changes in Earth’s rate of spin, the long timescale covered by ancient observations is an important asset. Approximately one million days, each marginally shorter than at present, have elapsed since the earliest reliable eclipse observations were made, about 700 bce. The contribution of individual small increments is summative. As a result, present-day computations of ancient eclipses that make no allowance for any increase in the length of the day may be as much as five or six hours ahead of the observed time of occurrence. In the case of total solar eclipses, the path of the Moon’s shadow across Earth’s surface may appear to be displaced by thousands of kilometres.

The technique of using ancient observations to investigate changes in the rate of Earth’s rotation is well illustrated by a total solar eclipse observed by Babylonian astronomers on a date corresponding to April 15, 136 bce. This event is recorded on two damaged tablets, a composite translation of which follows:

At 24 degrees after sunrise, there was a solar eclipse beginning on the southwest side. After 18 degrees it became total such that there was complete night. Venus, Mercury, and the normal stars were visible. Jupiter and Mars, which were in their period of disappearance, were visible in that eclipse. [The shadow] moved from southwest to northeast. [Time interval of] 35 degrees for obscuration and clearing up.

This is an exceptionally fine account of a total solar eclipse and is by far the best preserved from the ancient world. The Babylonians were able to detect a number of stars, as well as four planets, during the few minutes of darkness. Modern calculations confirm that Jupiter and Mars were too near the Sun to be observed under normal circumstances; Jupiter was very close to the solar disk.

As noted above, time intervals were expressed by the Babylonians in degrees, each equivalent to 4 minutes of time. Hence, the eclipse is recorded as beginning 96 minutes after sunrise (or about 7:10 am), becoming total 72 minutes later and lasting from start to finish for 140 minutes. Computations that make no allowance for changes in the length of the day displace the track of totality far to the west and imply that this eclipse was barely visible at Babylon, with as little as 15 percent of the Sun being covered. Furthermore, the computed time of onset is around noon rather than in the early morning—a difference of 3.4 hours. In order to best comply with the record, it is necessary to assume that the length of the day has increased by about 1/30 second in the intervening two millennia or so.

Numerous eclipses of both the Sun and the Moon were timed by the Babylonian astronomers with similar care, and analysis of the available records closely confirms the above result for the change in the length of the day. Although the timing device used is likely to have been of low precision, many eclipse observations were made fairly close to the reference moments of sunrise or sunset. For these the measured intervals would be so short that clock errors may be presumed to be small.

The many Arab and Chinese observations of both lunar and solar eclipses during the Middle Ages enable further variations in Earth’s spin to be traced. The following observations of the lunar eclipse of September 17, 1019, made by al-Bīrūnī at Ghazna attest to the quality of some of these more recent data:

When I observed it, the altitude of Capella [Alpha Aurigae] above the eastern horizon was slightly less than 60 degrees when the cut at the edge of the Full Moon had become visible; the altitude of Sirius [Alpha Canis Majoris] was [then] 17 degrees, that of Procyon [Alpha Canis Minoris] was 22 degrees and that of Aldebaran [Alpha Tauri] was 63 degrees, where all altitudes are measured from the eastern horizon.

All four stellar measurements are in agreement that the eclipse began at around 2:15 am, but calculations that make no allowance for any change in the length of the day indicate a time approximately 0.5 hour later. Observations such as these reveal that around 1000 ce the length of the day was about 1/65second shorter than at present.

Combining the various results obtained from analysis of ancient and medieval data, it is possible to show that over the last 2,700 years the rate of increase in the length of the day has varied markedly. This emphasizes the importance of nontidal effects in producing changes in the rate of Earth’s rotation period. In sum, the history of Earth’s rotation is extremely complex.