Colin J Humphreys and W Graeme Waddington report on the oldest recorded solar eclipse, a biblical reference which may be used to date precisely the reign of Ramesses the Great.

A puzzling event in The Bible that mentions both the Moon and the Sun can be interpreted as describing a solar eclipse. We have dated it to 30 October 1207 BC, making it possibly the oldest datable solar eclipse recorded. This enables us to refine the dates of certain Egyptian pharaohs, including Ramesses the Great. It also suggests that the expressions currently used for calculating changes in the Earth's rate of rotation can be reliably extended back 500 years, from 700 BC to 1200 BC.

In modern astronomy, solar eclipses are categorized into three types: total, annular and partial. In the ancient world, however, observers did not distinguish between total and annular solar eclipses. For example, the Han and later Chinese records indiscriminately apply the same expression chi (“total”) to both total and annular eclipses. On the other hand, the Chinese records do have a separate word for a partial solar eclipse (chin; Stephenson 1997). It is only when we get to the eclipse of 28 July AD 873, observed in Nishapur, Iran, that we have an unambiguously explicit statement of annularity from Al-Biruni.

In a total solar eclipse, the Moon covers the disc of the Sun with only an annulus of white light from the surrounding corona being visible, the level of illumination from which is roughly equivalent to that from a full Moon. In an annular eclipse, the silhouette of the Moon's disc is surrounded by a thin annulus of light from the uneclipsed Sun and the level of illumination on the Earth is roughly equivalent to dusk. In early times, both were called total and what was important was whether such eclipses happened or not.

To put our proposed ancient eclipse into context, we consider briefly the earliest recorded solar eclipses that have previously been suggested. A list of these is given by Espenak (2009). He gives four possible recorded solar eclipses before 1000 BC (the path of a total or annular eclipse is narrow, so the likelihood of ancient eclipse records occurring and surviving is small). His earliest suggested eclipse is ascribed to two Chinese observers, Ho and Hi. Legend says they were too drunk to see the eclipse and report it to the Emperor, so he had them executed. Espenak (2009) has calculated a possible eclipse for this time period and suggests it may have been an annular solar eclipse on 22 October 2137 BC. However, this well known story is most probably apocryphal and Stephenson (2008) does not even mention it in his detailed survey of ancient eclipses, so we rule this out as being a reliable eclipse record.

Clay tablet

The next oldest eclipse listed by Espenak (2009) was dated originally to 3 May 1375 BC (Stephenson 1970) and was thought to be a total solar eclipse recorded on a clay tablet found at Ugarit, in what is now Syria. However, a reanalysis gave a revised date of 5 March 1223 BC (de Jong & van Soldt 1989). Further analysis showed that the text on this tablet is best translated as: “During the six days of the (rituals of) the new moon of (the month of) Hiyyaru, the sun set, her gatekeeper (being) Rashep.” Since this text does not seem to refer to an eclipse at all, the eclipse interpretation has been firmly rejected (Pardee & Swerdlow 1993, Pardee 2002). We therefore rule out this suggested record of an eclipse.

The subsequent solar eclipse on Espenak's (2009) list is dated 5 June 1302 BC. This refers to Chinese writings on animal bones which have been translated as: “Three flames ate the Sun, and big stars were seen” (Pang et al. 2002). Stephenson (2008) has reanalysed this eclipse in detail, stating that this translation is not widely accepted and that other Chinese scholars believe the text refers to the weather. In addition, he considers that the technique Pang et al. (2002) used in revealing a date of c. 1300 BC for the supposed eclipse is “singularly unscientific” and concludes that this supposed eclipse record is “valueless” for astronomical purposes.

The only other eclipse record in the pre-1000 BC list of Espenak (2009) is a total solar eclipse dated 16 April 1178 BC, known as the Odyssey Eclipse on account of a passage in Homer's Odyssey, probably written in about 800 BC: “Ghosts … are going to Erebus beneath the dark; the Sun has perished from the sky, and an evil mist hovers over all.” Gainsford (2012) has made a detailed assessment of this text and concludes that it cannot plausibly refer to an eclipse. He states: “the passage refers to souls descending to Erebus (Hades) where notoriously the Sun does not shine.” Again, Stephenson (2008) does not even mention this supposed eclipse in his survey of ancient eclipses.

We therefore concur with Stephenson (2008) that, until now, there have been no reliable references to solar eclipses being observed before 1000 BC.

A possible observation

There is a possible reference to a solar eclipse in a puzzling passage in the biblical Old Testament book of Joshua. This records that, after Joshua had led the people of Israel into Canaan, he prayed: “Sun, stand still [Hebrew dôm] at Gibeon, and Moon, in the Valley of Aijalon.” The passage continues: “And the Sun stood still, and the Moon stopped [Hebrew 'amad], until the nation took vengeance on their enemies,” (Joshua 10:12–13, New Revised Standard Version [NRSV]). The locations of Gibeon and the Valley of Aijalon are shown in figure 1.

1

Map of ancient Canaan showing the route taken by the Israelites, starting at Gilgal, according to Joshua 10:9–10.

If these words are describing a real observation, then a major astronomical event was being reported (“There has been no day like it, before or since”, Joshua 10:14), but what does the text mean? The Hebrew word dôm means to be silent, dumb or still. The term 'amad is a broader word meaning to stop or stand. Modern English translations of this passage, such as the NRSV quoted above, have all followed the King James Authorized Version (KJAV) of The Bible, translated in 1611, and assumed that the Hebrew text means that the Sun and Moon stopped moving. However, a plausible alternative meaning is that the Sun and Moon stopped doing what they normally do: they stopped shining. In other words the text is referring to a solar eclipse, when the Sun stops shining. As a solar eclipse can only occur when the Moon is directly between the Earth and the Sun, the Moon itself is not visible and so it is not reflecting sunlight to the Earth – like the Sun, it has “stopped shining” as well.

The first person to suggest that Joshua 10:12–14 was referring to a solar eclipse seems to have been the linguist Robert Wilson (1918), who almost 100 years ago gave the following translation:

Be eclipsed, O sun, in Gibeon,

And the moon in the valley of Aijalon!

And the sun was eclipsed and the moon turned back, while the nation was avenged on its enemies.

Wilson claimed that in Babylonian cuneiform texts there are words with the same root as the Hebrew dôm that are used in Babylonian astronomical tablets in connection with eclipses, meaning “to be dark”. However, at that time, 100 years ago, it was not deemed possible to investigate this further because of the laborious nature of the calculations required (Russell 1918).

If the solar eclipse interpretation of this passage in Joshua is correct, then the text describes it as having been seen by the Israelites in Gibeon, in Canaan. Independent Egyptian evidence that the Israelites were in Canaan comes from the Merneptah Stele, a large inscribed granite block now housed in the Egyptian Museum in Cairo. The Egyptian Pharaoh Merneptah was the son of the well known Ramesses the Great (Ramesses II). The inscription on the Stele says it was carved in the fifth year of the reign of Merneptah and mentions a campaign in Canaan in which he defeated people of Israel. So the Israelites must have been in Canaan by Merneptah's fifth year.

The dates agreed by mainstream Egyptologists for the reign of Ramesses II are c. 1279–1213 BC, with his son Merneptah reigning from c. 1213–1203 BC (Shaw 2003, Horning et al. 2006, Kitchen 2013). These dates are subject to some uncertainty, with the latest possible dates for Ramesses II being 1270–1204 BC, and for Merneptah 1204–1194 BC (Kitchen 2013). The fifth year of Merneptah was therefore probably c. 1209/08 BC, with the latest possible date being 1200/1199 BC. Some other researchers, most notably including Rohl (1995), have proposed an alternative chronology for ancient Egypt in which these dates are advanced by several hundred years. Their “New Chronology” has achieved widespread publicity alongside widespread criticism from mainstream Egyptologists. In this “New Chronology”, the fifth year of Merneptah is 867 BC.

Sawyer (1972) followed up the suggestion of Wilson (1918) that Joshua 10:12–14 refers to a solar eclipse and considered the dates of all total solar eclipses visible from Gibeon between 1500 and 1050 BC (giving generous limits to the possible dates of the entry of Joshua into Canaan). He finds that there were only two such eclipses, on 19 August 1157 BC and on 30 September 1131 BC. However, both of these dates are significantly later than the latest possible date for Joshua to have entered Canaan, considered as the latest possible date for the fifth year of Merneptah, 1200/1199 BC. Historians and biblical scholars have therefore, to date, ruled out a solar eclipse interpretation of Joshua 10:12–14 (Walton 1994).

People in the ancient world did not distinguish between total and annular solar eclipses. It is not until as late as AD 1292, for instance, that we find a separate expression in the Chinese eclipse records to describe an annular eclipse; both of the annular eclipses of 7 August 198 BC and 27 July AD 306, for instance, were recorded by the Chinese as being total. We have therefore revisited the solar eclipse interpretation of Joshua 10:12–14 to see if there was an annular eclipse visible in the same time frame as was used by Sawyer.

In investigating the visibility of eclipses for this period we have used our own eclipse code, which conforms to the IAU 2006 recommendations (Hilton et al. 2006). The code uses our own fit of the French VSOP87/ELP2000-82b semi-analytical ephemerides to the JPL DE406 long-term integration. This fit required not only the adjustment of the secular terms to conform to the underlying basis of the JPL integration, but also the addition of a number of higher-order perturbation terms that were omitted from the French lunar ephemeris. At an epoch of 1000 BC, the r.m.s. deviation of the position of the Moon given by our code from that of DE406 is 0.3″ in ecliptic longitude and 0.2″ in ecliptic latitude. Thus, to all intents and purposes, our calculations may be said to have effectively used DE406. To facilitate the calculations we adopted the latest solution for the historical variations in the Earth's rotation (Stephenson et al. 2016); because these used a different ephemeris (DE432), due allowance has been made for the differing assumed lunar secular accelerations.

From our calculations we find that the only annular eclipse visible from Gibeon between 1500 and 1050 BC (using the same generous limits to the possible dates of entry of Joshua into Canaan as did Sawyer [1972]) was on 30 October 1207 BC, in the afternoon. Our calculated track of the annular eclipse of 30 October 1207 BC is shown in figure 2. This eclipse passed directly over the land of Canaan.

2

The path of the annular solar eclipse of 30 October 1207 BC, which passed directly over the land of Canaan in the afternoon. The shadow leaves the Earth's surface at sunset over modern day Iraq. The map is centred on Azekah, which is marked with a circle.

Solar eclipse of 1207 BC

According to the book of Joshua in the Old Testament, after an all-night march from Gilgal, the Israelites attacked the Amorites at Gibeon, they then pursued them to Azekah and then to Makkedah (figure 1). We have evidence from historical geography of where these places were: Gibeon was about 10 km northwest of Jerusalem, Azekah about 30 km southwest of Gibeon and Makkedah about 20 km south of Azekah (Notley & Rainey 2014). Because the eclipse occurred in the afternoon, it was probably seen from near Azekah, from where the partial eclipse would have started at 15:27 (local apparent time as given by a sundial), with annularity occurring between 16:48 and 16:53. The Sun would still have been partially eclipsed at sunset, which occurred at 17:38. During annularity, 86% of the solar disc's area was covered by the Moon.

An interesting feature of the Joshua text is the observation that it is stated that not only did the Sun stop (shining) but that the Moon also stopped (shining). As the Moon is in conjunction at the time of a solar eclipse it is effectively absent from the sky for a couple of days (it has “stopped shining”). As the Israelites used an observationally based lunar calendar they would have been well aware of this monthly period of lunar invisibility and so could have timed their surprise night-time attack at Gibeon to take advantage of the lack of natural night-time illumination at this time.

After reporting that the Sun stopped (shining), the book of Joshua states further that “The Sun did not hurry to set for about a whole day” (Joshua 10:13, NRSV), which has given rise to the term “Joshua's Long Day”. What did the writer mean? Figure 3 shows the level of illumination on the ground at Azekah during the annular eclipse and figure 4 shows the appearance of the Sun as viewed from Azekah at three-minute intervals.

3

The level of illumination on the ground at Azekah during the annular eclipse of 30 October 1207 BC, as a function of the local apparent time as given by a sundial. The thin line shows the normal illumination in the absence of an eclipse. The red line gives the illumination during the eclipse, showing the double-dusk effect.

4

The appearance of the Sun viewed from Azekah on 30 October 1207 BC at three-minute intervals. The Sun was still partially eclipsed at sunset.

All ancient civilizations would have been accustomed to the Sun going down in the afternoon, leading to daylight turning into dusk, and then turning into night. However, on this occasion, in the afternoon the light from the Sun on Canaan started decreasing from its normal level at about 15:30 until at about 16:50 it was approximately ten times less intense than normal and dusk set in (notice that figure 3 is plotted on a logarithmic scale to match the approximate response of the human visual system). However, by around 17:10 the level of illumination would have been somewhat restored before dusk fell again and then the Sun finally set at about 17:38. In pre-scientific cultures such an unexpected deviation from normal behaviour on the part of the Sun could only inspire awe and the perceived change in the ambient light level would naturally lend itself to description in terms of the normal order of things – namely, dusk. What the Israelites would have witnessed was a double dusk. To the awe-inspired Israelites of 1207 BC, the amazing spectacle in the sky would have appeared to be long and drawn-out; the reaction to such events tends to be exaggerated, particularly with regard to perceived duration. For example, the solar eclipse of 18 July AD 1860 was observed in Sudan by Mahmoud Bey who reported: “To everyone the two minutes of the eclipse were like two hours … Several people whom I questioned after the eclipse regarding the duration of totality replied that it had lasted for two hours” (see Faye 1860).

In attempting to describe this double dusk it is only natural that the Israelites would have done so in terms of their normal experience of the diurnal cycle. Although aware that on this occasion the time interval between the two dusks was less than the normal day, the book of Joshua records “about a whole day” (NRSV) for this period of time. In fact the Hebrew text here is “like a whole day”, the preposition like also means as, and so the phrase can mean “as on a whole day” (Millard, private communication). Thus the analogy being employed is one of following the diurnal rise and fall of the ground illumination.

The appearance of the annular eclipse of 30 October 1207 BC we are considering is shown in figure 4. Both before and after annularity, the eclipse takes on the appearance of a crescent, mimicking the form of the Moon around both the end and beginning of a lunar month. This changing appearance of the Sun may well have brought to mind the period of lunar invisibility at the changing of the lunar month when the Moon stops shining. So the description in the book of Joshua of a celestial event in which both the Sun and the Moon stopped shining is consistent with the observation by an ancient Israelite layman of an annular solar eclipse.

Historical implications

Pre-telescopic eclipse records are of considerable chronological interest; the total solar eclipse of 15 June 763 BC, for example, was recorded in Assyrian records (Rawlinson 1867) and is now used as a key fixed point to date Assyrian kings objectively over most of the surrounding three centuries. If our solar eclipse interpretation of Joshua chapter 10 is accepted, it has consequences for the chronology of the ancient world.

As stated above, the Israel Stele of Merneptah refers to his confrontation with people of Israel. The Stele is dated to Merneptah's year 5, which was the year of his most recent victory against the Libyans. The confrontation with Israel probably occurred in his year 2 to 4 (Kitchen 2006), so 1207 BC is probably year 2, 3 or 4 of Merneptah. If accepted, this would conclusively rule out the “New Chronology” of Rohl (1995) and others for ancient Egyptian Pharaohs. It also enables us to revise by a few years the mainstream Egyptian chronology.

The standard Egyptian chronology gives the dates of Merneptah as 1213–1203 BC, with the latest possible dates being 1204–1194 BC. If the confrontation with Israel was in year 4 of the reign of Merneptah, then his reign would have started in 1211/1210 BC, if in year 3, 1210/1209 BC, and if in year 2, 1209/1208 BC. Hence we can pinpoint the first year of Merneptah as 1210/1209±1 year. As the length of his reign is known from Egyptian texts, his reign would have lasted between 1210 BC and 1200 BC (with ±1 year in each case). The dates of the previous and subsequent pharaohs can be similarly adjusted; for example, the reign-dates of Ramesses II (Ramesses the Great) would be 1276–1210 BC ±1 year.

The Earth's rotation

Pre-telescopic observations of ancient eclipses are of considerable scientific value in studying the long-term variations in the rate of rotation of the Earth. However, only a small number of eclipse records are available from before 700 BC.

Such variations are mainly produced by the lunar and solar tides, but non-tidal mechanisms are also significant (Stephenson 1997). The accumulated clock error arising from changes in the Earth's rate of rotation is known as ΔT. Values of ΔT before AD 1600 pre-date the telescope and are based on historical records of naked-eye observations of eclipses going back to 700 BC. These observations show that the rate of rotation of the Earth is slowing at an average value over the last 2700 years of about 1.8 ms per day per century (about one minute per century) (Stephenson 1997).

The theoretical tidal contribution to ΔT is a constant deceleration of the Earth's rotation of about 2.4 ms per day per century, hence there is about −0.6 ms per day per century to be explained. As Stephenson and Morrison have shown (Stephenson et al. 2016, Stephenson & Morrison 1995, Morrison & Stephenson 2001), this discrepancy is consistent with post-glacial rebound following the last Ice Age, changing the Earth's moment of inertia as the polar ice caps shrank. In addition there appears to be a pseudo-periodic contribution to the length of day with a period of about 1500 years, the nature and reality of which is uncertain (Stephenson et al. 2016, Huber 2006). Fortuitously, the phase of this proposed periodic contribution is such that at 1207 BC its contribution is effectively zero (Stephenson et al. 2016).

In our calculations we have extrapolated back 500 years to 1207 BC expressions for ΔT based on observations extending back to only 700 BC. Although it may be considered unwise to extrapolate these back prior to 1000 BC (Morrison & Stephenson 2004), Stephenson (2008) states that his solution for the long-term variation of ΔT “should provide a useful first approximation to ΔT as far back as 1500 BC”. We also note that paleorotation data from fossil bivalves going back 640 million years (Varga et al. 1998) give an average rate of change of the length of day in agreement with the astronomical data going back 2700 years.

Our analysis of the text of the book of Joshua suggests there was an annular eclipse seen over Canaan. Calculations, using the best available value of ΔT, are consistent with this and date the eclipse to 30 October 1207 BC. If this is accepted, it suggests that the long-term trend of ΔT going back to 700 BC can be reliably extended a further 500 years, back to 1207 BC.

Conclusion

A reinterpretation of a puzzling passage in the Old Testament book of Joshua suggests that a solar eclipse was being reported. Calculations show that this event could be the annular solar eclipse of 30 October 1207 BC. If accepted, this appears to be the oldest solar eclipse recorded. When combined with Egyptian records, this eclipse enables us to hone the most accurate dates available for the reign of the famous Egyptian pharaoh Ramesses the Great to be 1276–1210 BC ±1 year. This work further suggests that the expressions currently used for calculating ΔT, the accumulated clock error due to changes in the Earth's rate of rotation, can be extended back 500 years from 700 BC to 1200 BC.

ACKNOWLEDGMENTS

The authors thank Alan Millard, Emeritus Professor of Hebrew and Ancient Semitic Languages at the University of Liverpool, for his help with understanding ancient texts, and Colin Bell at Tyndale House, Cambridge, for drawing figure 1. We also thank Michael Treacy, Professor of Physics at Arizona State University, for stimulating discussions.

REFERENCES

de Jong
T
&
van Soldt
W H
1989
Nature
338
238

Espenak
F
2009
Solar eclipses of historical interest
NASA Eclipse
Web Site http://eclipse.gsfc.nasa.gov/SEhistory/SEhistory.html

Faye
H A E A
1860
Comptes rendus hebdomadaires des séances de l'Académie des Sciences
51
680

Gainsford
P
2012
Trans. Amer. Phil. Assoc.
142
1

Hilton
J L
et al. 
2006
Celestial Mechanics and Dynamical Astronomy
94
351

Horning
E
et al.  (eds)
2006
Ancient Egyptian Chronology
(Brill)

Huber
P J
2006
J. Geod.
80
283

Kitchen
K A
2006
On the Reliability of the Old Testament
(
Eerdmans
)

Kitchen
K A
2013
Establishing chronology in pharaonic Egypt and the Ancient Near East
, in
Radiocarbon and the Chronologies of Ancient Egypt
Shortland
A J
&
Ramsey
C B
(eds) (
Oxbow Books
)

Morrison
L V
&
Stephenson
F R
2001
J. Geodynamics
32
247

Morrison
L V
&
Stephenson
F R
2004
J. Hist. Astr.
35
327

Notley
R S
&
Rainey
A F
2014
The Sacred Bridge: Carta's Atlas of the Biblical World, Second Edition
(
Carta
)

Pang
K D
et al. 
2002
Astronomical dating and statistical analysis of ancient eclipse data
, in
History of Oriental Astronomy
Ansari
S M R
(ed.) (
Kluwer Academic
)
95

Pardee
D
2002
Ritual and Cult at Ugarit
(
Society of Biblical Literature
)

Pardee
D
&
Swerdlow
N
1993
Nature
363
406

Rawlinson
H C
1867
The Assyrian Canon Verified by the Record of a Solar Eclipse, B.C. 763
Athenaeum #2064 660 (18 May)

Rohl
D
1995
A Test of Time (Century)
.
Pharaohs and Kings (Crown)

Russell
H N
1918
The standing still of the Sun (Joshua 10:12–14)
,
Princeton Theological Review
16
103

Sawyer
J F A
1972
Palestine Exploration Quarterly
104
139

Shaw
I
(ed.)
2003
The Oxford History of Ancient Egypt
(
Oxford
)

Stephenson
F R
1970
Nature
228
651

Stephenson
F R
1997
Historical Eclipses and Earth's Rotation
(
Cambridge
)

Stephenson
F R
2008
J. Hist. Astr.
39
229

Stephenson
F R
&
Morrison
L V
1995
Phil. Trans. Roy. Soc. A
351
165

Stephenson
F R
et al. 
2016
Proc. R. Soc. A
472
20160404

Varga
P
et al. 
1998
J. Geodynamics
25
61

Walton
J H
1994
Joshua 10:12–15 and Mesopotamian celestial omen texts
, in
Faith, Tradition and History
Millard
A R
,
Hoffmeier
J K
&
Baker
D W
(eds) (
Eisenbrauns
)

Wilson
R D
1918
Princeton Theological Review
16
46