杏吧原创

AT 8.45 on the morning of 15 April 136 BC, Babylon was plunged into darkness
when the Moon passed in front of the Sun. An astrologer, who recorded the
details in cuneiform characters on a clay tablet, wrote: 鈥淎t 24 degrees after
sunrise鈥攁 solar eclipse. When it began on the southwest side, Venus,
Mercury and the normal stars were visible. Jupiter and Mars, which were in their
period of disappearance, became visible. The Sun threw off the shadow from
southwest to northeast.鈥�

There is no reason to doubt this ancient account of a total eclipse. But
consider this. If present-day astronomers use a computer to run the movements of
the Earth, Moon and Sun backwards from their present positions, like a movie in
reverse, they find something very odd. The total eclipse of 15 April 136 BC
should not have been visible from Babylon at all. The zone of totality should
have passed over the Spanish island of Mallorca, 48.8 degrees west of
Babylon鈥攁 difference of more than one-eighth of a complete rotation of the
Earth, or 3.25 hours (see Diagram).

How can this be? The only explanation is that the planet鈥檚 rotation has
slowed since 136 BC, making the day longer. Of course, there are many other
records of the ancients observing cosmic events, from supernovas to comets, but
the value of these sightings to modern science is limited. Reports of eclipses,
however, are in a class of their own. If the Earth has accumulated a change in
orientation equivalent to an eighth of a turn in just over 2000 years, then we
can infer that the day has lengthened by an average of a few milliseconds a
century. This is an extraordinarily precise figure to deduce from historical
records. In fact, it is without precedent.

The eclipse records are unique in another way too. Richard Stephenson of the
University of Durham and Leslie Morrison, formerly of the Royal Greenwich
Observatory in Cambridge, are using the sightings in a long-term project to
discern subtle changes in the Earth鈥檚 spin. 鈥淪uch records are the only way known
to us at present of measuring the actual change in the rotation of the Earth
over the course of recorded history,鈥� says Stephenson.

The key to the extraordinary accuracy of the technique is the fact that, at
this moment in the Solar System鈥檚 history, the Moon and Sun appear the same size
when viewed from Earth. This generates an eclipse 鈥渢rack鈥� that is at most 250
kilometres wide, making total eclipses at any spot on Earth very rare
indeed.

By using the trick of running the motions of Earth, Moon and Sun backwards,
Stephenson and Morrison can calculate the time of an eclipse to the second. They
can also deduce where the eclipse should have been visible if the length of day
has always been the same as it is now. Comparing the calculated eclipse track
with where the eclipse was seen, reveals how much the day has lengthened. The
beauty of their method is that it relies solely on knowing an observer鈥檚
location鈥攁nd which eclipse they saw. 鈥淵ou don鈥檛 even need a precise date
to identify it,鈥� says Stephenson. 鈥淧lus or minus twenty years is usually good
别苍辞耻驳丑.鈥�

Stephenson and Morrison have so far accumulated about 300 eclipse reports,
both lunar and solar, stretching back as far as 700 BC. Most come from four
civilisations: the Babylonian, Chinese, European and Arab. The Babylonian
records were probably made by astrologers who used them to predict future
celestial events鈥攆or horoscopes, perhaps. They cover the period between
700 BC and AD 75. The Babylonians also recorded lunar eclipses and, remarkably,
timed all eclipses to the nearest four minutes.

So extensive were their records that the Babylonians even spotted that
eclipses with similar characteristics recur every 18 years and 11鈪� days, known
today as the saros cycle. Most of these records, written on clay tablets, were
unearthed in the 19th century by peasants looking for bricks and sold to antique
collectors in Baghdad. From there they found their way to the British Museum,
which is now the repository of all the known tablets.

Chinese records of eclipses were made by civil servants paid to watch for
omens such as new stars, comets and meteor showers. Many eclipses are timed to
an accuracy of about 15 minutes. Reliable Chinese records extend from the 8th
century BC to modern times. They include the earliest continuous sequence of
ancient solar eclipses鈥�36 including three total eclipses鈥攂etween 720
BC and 480 BC.

By contrast with the Chinese and Babylonian records, ancient European records
are scattered. Occasional references to eclipses were made by Greek and Roman
writers such as Herodotus and Plutarch. The Greek poet Archilochus describes an
eclipse in a poem from 650 BC, while Ptolemy included a collection of lunar
eclipses dating back to 720 BC (probably from Babylonian records) in his
Almagest.

Medieval witnesses

Later European accounts of eclipses are found in the medieval chronicles,
which were a major literary form between 800 and 1400. 鈥淭hey were written in
diary form by laymen, often in monasteries, and they provide some splendid
descriptions with very accurate dates,鈥� says Stephenson. The chronicler Leo, for
example, watched the total eclipse of 22 December 968 from Constantinople, and
wrote: 鈥淎t the winter solstice there was an eclipse of the Sun such as has never
happened apart from that which was brought on the Earth at the Passion of our
Lord on account of the folly of the Jews鈥� Darkness fell upon the Earth and all
the brighter stars revealed themselves. Everyone could see the disc of the Sun
without brightness, deprived of light, and a certain dull and feeble glow, like
a narrow headband.鈥�

According to Stephenson, mention of a 鈥渉eadband鈥� is the earliest reference to
the Sun鈥檚 outer atmosphere, or 鈥渃orona鈥�, that is definitely datable. It is even
possible to deduce from Leo鈥檚 description of the corona as 鈥渘arrow鈥� that the Sun
was near the minimum of its cycle of magnetic activity.

The Arabs also produced a fair number of chronicles during medieval times. On
11 April 1176, for instance, a 16-year-old called Ibn al-Athir witnessed a total
eclipse of the Sun at Cizre, a Turkish town on the frontier with Syria.
Recalling the event, many years later, he wrote: 鈥淚n this year the Sun was
eclipsed totally and the Earth was in darkness so that it was like a dark night
and the stars appeared. That was the forenoon of Friday the 29th of the month of
Ramadan at Jazirat Ibn 鈥楿mar, when I was young and in the company of my
arithmetic teacher. When I saw it I was very much afraid. I held on to him and
my heart was strengthened.鈥�

But the recording of eclipses in the Arab world was not left just to
chroniclers. Between AD 800 and 1000, astronomers in Baghdad and Cairo noted the
times of eclipses to within 5 minutes.

Altogether, the records that Stephenson and Morrison have found show that in
500 BC the day was about 50 milliseconds shorter than it is today. When all the
data are plotted, one effect is clearly visible and that is the linear increase
in the length of the day because of the tidal interaction between Earth and Moon
(and to a lesser extent between Earth and Sun). Basically, differences in the
Moon鈥檚 gravity stretch the Earth to make a 鈥渂ulge鈥�, most noticeably in the
oceans, which the Moon then tugs on as the Earth rotates. The net effect is that
the Earth鈥檚 spin gradually slows, while the Moon recedes from the Earth by about
3.7 centimetres a year.

Calculations show that the Moon鈥檚 tidal influence on Earth should increase
the length of day by an average 2.3 milliseconds a century. But this conflicts
with Stephenson and Morrison鈥檚 calculations. They found an average increase over
the past 2500 years of only 1.7 milliseconds a century. Stephenson says the
difference is down to another large influence on the Earth鈥檚 spin. 鈥淚t is
connected to the disappearance of the ice sheets from most of the planet 10 000
years ago,鈥� he says.

The enormous weight of ice that built up during the last ice age flattened
the planet slightly at the poles. When the ice began to melt, the land began to
rise slowly. This process of 鈥減ost-glacial rebound鈥� is still going on today. Its
effect is to make the Earth slimmer round the equator. And, like an ice skater
pulling in his or her arms, the planet spins faster, shortening the day by
between 0.5 and 0.6 milliseconds a century.

So everything seems to fit. Taking into account the lengthening of the day
caused by the Moon鈥檚 gravity and the shortening effect of post-glacial rebound,
the day should be growing longer by about 1.7 milliseconds a century, which is
just what Stephenson and Morrison have found. 鈥淚t could, of course, just be
fortuitous,鈥� says Stephenson. 鈥淏ut we don鈥檛 think so.鈥�

Morrison points out that 1.7 milliseconds a century is very much an average.
鈥淭he figure we see in records can go as low as 1.4 milliseconds per century and
as high as 2.0 milliseconds per century,鈥� he says. 鈥淚n fact, it oscillates with
a period of about 1000 years.鈥�

This fluctuation has been brought into sharper focus since the arrival of the
telescope around 1600. Not long after this, total eclipses ceased to be the
prime record of day length for present-day astronomers. Far better are
鈥渙ccultations鈥�, when known stars disappear behind objects such as the Moon.
Observations of occultations since about 1650 reveal fluctuations in the Earth鈥檚
spin of the same magnitude as the 1000-year oscillation but on timescales of
only decades鈥攖hese have been dubbed decade fluctuations. 鈥淚t seems that
the 1000-year cycle is merely the envelope of the decade fluctuations,鈥� says
Morrison. 鈥淚t鈥檚 what you see when you have insufficient resolution.鈥�

The cause of the long-term oscillation is still a mystery, although
Stephenson thinks it could be connected with changes in sea level, which subtly
change the shape of the Earth. Another possibility, says Morrison, is that it
has something to do with changes in the magnetic field in the Earth鈥檚 core which
affect the motion of the surrounding mantle. To date, however, nobody knows what
would cause such changes or how this 鈥渃ore-mantle coupling鈥� would work.

The task of explaining fluctuations in the Earth鈥檚 spin is one for theorists.
Stephenson and Morrison see their job as giving the theorists something to think
about. 鈥淭he whole point of the work is to discover new and unexpected influences
on the Earth鈥檚 spin,鈥� says Stephenson.

In their search for such influences, the two are looking for yet more
historical data. They would dearly like to extend their record of eclipses back
before 700 BC. The Egyptian civilisation extends back for several thousand years
before the Babylonian, and the ancient Egyptians must have seen many eclipses.
鈥淭he problem is finding the references and then interpreting them,鈥� says
Morrison. 鈥淔or instance, they may have equated such a dramatic celestial event
with a dramatic terrestrial event such as the death of a king.鈥�

Stephenson and Morrison also have gaps in their data between 100 BC and AD
500, and between 1300 and 1600. They hope that Chinese records of occultations
might help them to fill in the earlier gap, while records from early European
astronomers could plug the later gap. The astronomers think they have found just
about all the useful European data, although it can be difficult to get at. 鈥淭he
Vatican Library only allows you to look at three manuscripts a day, which is a
bit of a problem when you鈥檙e simply looking for the odd reference,鈥� says
Stephenson.

Arab records are even more difficult to find. Stephenson believes there may
be vast numbers of manuscripts lying uncatalogued in obscure libraries. As for
Babylonian tablets, Stephenson speculates that only 10 per cent of what was
recorded has ever been found. 鈥淭he other 90 per cent may have been destroyed,鈥�
says Stephenson. 鈥淭hen again, they may still be in the ground. We live in
丑辞辫别.鈥�