Beruflich Dokumente
Kultur Dokumente
Lecture held at the Research Center for Theory and History of science of the
University of West Bohemia in Pilsen, 3rd March 2014, and at the Institute of
Philosophy of the Czech Academy of Sciences in Prague, 6th March 2014
Several sources, of which Herodotus is the most famous, report that Thales predicted an
eclipse of the sun. We may ask:
1) did he really predict an eclipse?
2) if so, which one did he predict?
3) if so, how did he do it?
The answers to the first two questions will appear to follow from the answer to the third one.
If we are able to discover which method Thales could have used, it follows immediately that
he predicted an eclipse and also which one it was he foretold.
We will answer these three questions: yes, he did predict a solar eclipse; it was the eclipse of
28 May 585 BC, and he did it by making a brilliant mistake. Here is the path of the eclipse of
28 May 585 BC:
As you see, Miletus lies just outside the path of the full eclipse, which means that what Thales
saw was actually a great but partial eclipse:
The solar
eclipse of
25 May 585
BC, seen
from
Miletus
2
But let us hear first some sceptical voices.
The main source, Herodotus, says that Thales foretold the eclipse, fixing as its term the sixth
year of the battle between the Medes and the Lydians. This can hardly be called a prediction.
Dmitri Panchenko rightly remarked: If one can predict an eclipse at all, one can predict it to
the day. Herodotus remark shows, how little he understood of eclipses and their prediction.
The most disappointing is the search for a method that Thales could have used for his
prediction, the search for some regularity in the occurrences of eclipses. It has been argued
that Thales must have had knowledge of ancient Babylonian wisdom about cycles of solar
eclipses.
In 1969 Willy Hartner examined fifteen possible cycles, of which the Saros (a cycle of 223
lunar months or lunations) and the Exeligmos (or triple Saros of 669 lunar months) are the
most important, but his conclusion was that none of them could have been a base for a
prediction of the eclipse of 28 May 585 BC.1
Dmitri Panchenko tried the Exeligmos for the eclipse of 21 September 582 AD.2
The solar
eclipse of
21
September
582 BC,
seen from
Miletus
Panchenko calls this eclipse impressive, but in 1997 Stephenson and Fatoohi, two experts
on eclipses, pointed out very clearly that this eclipse may very well have passed unnoticed.3
And they are right, as I could observe myself at the eclipse of 11 August 1999 AD, which was
in Holland (actually Maastricht) even bigger (0.95) than the eclipse of 21 September 582 BC,
which was 0.85 at Miletus.
The solar
eclipse of
11 August
1999 AD,
seen from
Maastricht
3
However, she based her analysis on eclipses, visible at Ninive instead of Miletus, and
her list contains some other dubious points. When we make the list for Miletus it appears that
no more than 5 out of the 22 lunar eclipses took place 23 1/2 month before a solar eclipse.
This is too small a base for a prediction that is more than a guess, as Hartner already
remarked.
Moreover, in Thales time even the Babylonian astronomers were not able to predict
eclipses. The only thing they were capable of was to find out the dates of possible eclipses,
based on a very simple rule of thumb. Of this rule of thumb we will come to speak later.
Our conclusion can be that Thales did not depend on ancient wisdom, whether
Babylonian or not.
It could be asked, whether a prediction of an eclipse does not presuppose a knowledge of the
true mechanism of eclipses. However, it must be stressed that the prediction of an eclipse does
not presuppose any knowledge whatsoever of what a solar eclipse really is.
Even Anaximander was not acquainted with this, for he explained eclipses as the partially or
totally closing of the apertures in the celestial wheel of the sun or the moon. In other words:
Thales did not need to know that an eclipse of the sun occurs when the sun is blocked by the
moon.
The only thing that matters that he observed eclipses when they occurred and tried to
find some regularity in their occurrences. These regularities Thales could have noticed
without having the faintest idea why they occurred.
Our conclusion can be that Thales did not depend on knowledge about the true
mechanism of eclipses.
At best, the sceptics say, Thales made a lucky guess. I will argue that his guess was brilliant
and based on an observed regularity.
From Willy Hartner I took the idea of looking at the statistical material at Thales
disposal, as he called it. Let us suppose, then, that Thales knew nothing about alleged
Babylonian wisdom, but just started observing and registrating eclipses of the sun and the
moon during his lifetime. Fortunately, he lived in a time in which relatively many eclipses
were visible at Miletus.
The first thing he could have discovered if this was not already common knowledge was
that both lunar and solar eclipses can be partial or full.
The second thing he could have discovered if this was not already common knowledge
was that lunar eclipses occur at full moon and solar eclipses at new moon.
This was helpful, because it meant that he did not need to look out for a possible
eclipse every day and night, but only twice a month. Mark that whenever I speak of months,
lunar months or lunations are meant, from one new moon until the next, about 29 1/2 days.
The third thing he could have observed was that the dates of possible solar eclipses
always fall within the same month as possible lunar eclipses. This will become more clear
when we will see the table with possible and real eclipses during Thales lifetime.
The fourth thing he could have learnt after some years of observation, was that not
every full or new moon an eclipse occurred, but that they came at longer intervals, of
multiples of six months, sometimes minus one. This is the rule of thumb the Babylonians of
his time also were acquainted with.
This rule presents the data of possible eclipses, days on which it is worth while to look
at the heavens. As you can see, actually the rule has a certain rhythm or pattern, which repeats
itself after 38 lunar months:6
This knowledge made observation still easier, because he had to observe the sky only twice a
year.
Together this is the knowledge that Thales could have been gathered simply from a
careful observation of the heavens during his lifetime. When put into a table, it looks like this:
YEAR
B.C.
lunar ()
and
solar
eclipses
( = full,
= partial)
visible at
Miletus
from 610586 B.C.
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
609
27
608
4
606
24
605
13
31
603
21
10
601
598
595
22
12
593
30
592
591
16
588
587
13
19
8
26
589
586
17
6
590
30
25
4
12
15
4
22
25
13
3
23
13
25
13
2
22
11
23
21
11
12
1
23
11
19
9
21
10
30
17
7
9
28
18
7
17
6
25
16
25
15
4
31
19
28
27
13
2
20
19
16
13
30
27
26
16
24
11
20
28
19
21
30
17
25
14
30
18
27
18
11
1
30
22
11
18
29
22
13
3
14
23
15
4
31
25
3
22
20
18
16
5
25
15
5
15
29
27
17
7
26
17
6
28
25
15
12 11
10
20
26
23
20
12
21
18
27
17
7
26
14
3
28
1 31
13
22
13
10
20
4
24
14
24
22
30
5
25
16
23
20
10
16
16
6
27
17
28
17
7
DEC
28
18
28
18
28
14
NOV
30
19
29
19
11
24
15
22
12
29
18
7
12
21
10
17
6
24
13
21
10
12
7
27
25
13
26
16
5
20
29
26
17
6
23
15
19
9
28
16
25
597
594
10
10
22
21
29
21
27
10
12
1
22
31
19
17
6
12
23
13
3
23
24
14
3
22
25
15
5
28
600
26
12
604
596
24
15
607
599
OCT
610
602
SEP
The cells of the months in which an eclipse of the moon or the sun was visible at Miletus are
colored red. The numbers in the cells of the months are the data of new moon.
I took the idea of this table from Joachim Herrmann, who in 1969 drew a similar one
for the years 1968 to 1990 AD.7 As we see, in such a table the rule of thumb expresses itself
as a series of strings or garlands through the calendar. This is even better visible when the
table is folded into a cylinder.
5
However, the extrapolation of this table will lead to no more than the fixation of two
possible data for solar eclipses in 585 BC: one at new moon in May and one at new moon in
November. For a real prediction, more is needed.
Let us suppose that Thales looked more carefully into the data he had gathered during
his lifetime. Then he could have noticed a curious regularity in the series of the last eight
successive eclipses of the sun before the year 585 BC. I will show you them in a table:
After how many months do you think the next
solar eclipse is to be expected?
I left out eclipses amaller than 0.50, because they could easily have escaped observation,
unless Thales had a special reason to expect one. If you were Thales, what would you have
put as the next item in the last column?
YES !!
And which date do you think will appear in the last row?
YES !!
Hartners mistake
28 June 596
0.35
12
23
From Dmitri Panchenko I took the idea of looking for apparent regularities, and more
in particular the intervals of 17 and 18 months. In 1994 he wrote; an astronomer (like Thales)
would naturally have searched for () eclipses with the same interval. And also he wrote:
the eclipse of 28 May 585 was separated from the two previous eclipses by intervals of 18
and 17 lunations respectively. But this sequence had taken place once before! The eclipse of
7
30 July 607 was also separated by intervals of 18 and 17 lunations from the two preceding
eclipses.
Why didnt Panchenko not discover the other interval of 17 and 18 months? Because
he blindly copied Hartners mistake and put the number 23 instead of 17 at the eclipse of 9
May 594:
23
Because of this mistake Panchenko favored the eclipse of 21 September 582 BC, as we
already saw.
The only thing I had to do was to correct Hartnerss and Panchenkos mistake.
And now for the last question: Did Thales really predict the solar eclipse? The answer
must be: no. The regularity he saw was only a lucky coincidence. When we look at eclipses
later than the year 585 BC, no such regularity as the 17 and 18 lunations occurs.
This explains why Thales predicted only one eclipse, and also why others could not
copy his method. After all, his preciction was a lucky guess, although it was based on
thorough research and careful observation. His prediction was an example of the not
uncommon situation in science that a right conclusion is based on a false presupposition. Such
mistakes happen in science and their exposure brings science further.
The End
Willy Hartner, Eclipse Periods and Thales Prediction of a Solar Eclipse. Centaurus 14 (1969), 60-71.
Dmitri Panchenko, Thaless Prediction of a Solar Eclipse. Journal for the History of Astronomy 25 (1994),
275-88.
3
Francis R. Stephenson & Louay J. Fatoohi, Thales Prediction of a Solar Eclipse, Journal for the History of
Astronomy 28 (1997), 279-82.
4
Patricia F. OGrady, Thales of Miletus. Aldershot: Ashgate 2002.
5
Bartel L. Van der Waerden, Science Awakening II: The Birth of Astronomy. Leiden: Brill 1974, 122, n. 1.
6
Cf. John M. Steele, Eclipse Prediction in Mesopotamia. Archive for the History of Exact Sciences 54 (2000),
421-54.
7
Joachim Herrmann, Gesetze des Weltalls, Stuttgart: Franckhsche Verlagshandlung , W. Keller & Co. 1969.