[Small Kitora Painting]

Kitora Kofun

A Detailed Astronomical Star Chart in an Ancient Japanese Tomb

By Steve Renshaw and Saori Ihara

Revised September, 2002

As mentioned in the supplement to our article on Takamatsu Zuka Kofun , The Controversy Continues, historians and archaeologists have long thought that more tombs nearby might contain similar "star paintings". One such tomb, located about 1 kilometer to the South of Takamatsu Zuka Kofun and named Kitora Kofun has now been explored. The tomb has not been actually entered but rather probed with sub-miniature camera. Archaeological dating places its construction within the same Asuka period (7th to early 8th centuries) as Takamatsu Zuka Kofun. While there are some remarkable similarities, there are also some anomalous differences in the paintings of the two tombs.

[Kitora from South]
Kitora Kofun (Kitora Tumulus) seen from South Looking North. (Photo by Steve Renshaw)

[Kitora from East]
Kitora Kofun from the East Side. Only the front of the tomb mound can be seen. The actual tomb is located in the lower (larger) mound directly under the upper (smaller) mound and is aligned North/South. (Photo by Steve Renshaw)

Composite of Images Showing the Inside of Kitora Kofun. (From Research Report of Cultural Heritage in Asuka Village Vol. 3; Scientific Research on Kitora Tumulus)

In 1983, a probe was inserted into Kitora Kofun from the South, and an indistinct image of Genbu (The Black Tortoise of the North) was found on the opposite (North) wall. Excited by the possibility of finding another set of tomb paintings similar to those in Takamatsu Zuka Kofun, plans were developed for further exploration. On March 5, 1998, under the direction of NHK and Tokai University, a probe with greater flexibility was again inserted into the tomb. Again, Genbu was found on the opposite wall; the following day, the distinct outline of Byakko (The White Tiger of the West) was seen on the West wall and Seiryuu (The Azure Dragon of the East) was found on the eastern wall of the tomb. As in Takamatsu Zuka Kofun, the ceiling was found to contain paintings of stars and constellations; several Sei Shuku (Moon Stations) were clearly distinct.

Composite of Images Showing the Ceiling of Kitora Kofun with Star Chart. (From Research Report of Cultural Heritage in Asuka Village Vol. 3; Scientific Research on Kitora Tumulus)

The tomb was again probed in December of 2001 with a larger digital camera. At this time, Suzaku (The Red Bird of the South) was found, such image having been completely destroyed in Takamatsu Zuka Kofun. While there are noticeable stains, paintings of the animals of cardinal directions are all intact and appear to be in somewhat better condition than those of the tomb to the north (in which some paintings appear to have been defaced in ancient times) The animals seem to have been painted in a somewhat freer style than those which are found in Takamatsu Zuka Kofun. However, the pattern used by the tomb artist appears to have been virtually identical. Still unexplained, the painting of the "white tiger" is reversed, unlike that of virtually every other tomb representation found in China, Korea, or Japan (See Krupp, 1983).

The Red Bird of the South (Suzaku) [From National Research Institute for Cultural Properties, Nara]

[Byakko Comparison]
Comparison of Paintings of Byakko in Kitora (left) and Takamatsu Zuka (right) tombs. (From Research Report of Cultural Heritage in Asuka Village Vol. 3; Scientific Research on Kitora Tumulus)

More stars appear in the Kitora tomb paintings. Unlike Takamatsu Zuka Kofun where moon stations are painted in a "square" surrounding the central north pole region, the Kitora Kofun astronomical star chart includes many other constellations, an inner circle marking stars which were seen to not set in the North, circles marking the celestial equator and ecliptic, and an outer circle marking the horizon line of the "observer". Despite initial news reports following the 1998 probe, a discolored "line" through the painting did not turn out to be a representation of the Milky Way; a delineation of such was not found. Like Takamatsu Zuka Kofun, the tomb appears to have been plundered of any "treasure" it may have held. However, the probe of 2001 showed depicted stars in Kitora Kofun to be painted in gold leaf as they were in the earlier explored tomb.

Big Dipper
The "Big Dipper" with stars in gold leaf (Note Mizar as a "double" star). [From National Research Institute for Cultural Properties, Nara]

Research on the Astronomy of Kitora Kofun

The Asuka Board of Education (see References below) has now published a research report on findings related to Kitora Kofun. Along with research in archaeological and other aspects of the tomb, this report also includes reports of work conducted by Professors Kazuhiko Miyajima and Keiji Yamada on astronomical aspects. While having provided information in some public forums, Dr. Miyajima made a formal presentation of academic research related to the astronomy of Kitora Kofun at a special session of the Astronomical Society of Japan in Fukuoka on October 10, 1999 at Kyushu University. In our presentation and analysis here, we will use material from both Miyajima's lecture and the Asuka Research Report.

[Miyajima 1]
Professor Kazuhiko Miyajima presents findings related to Kitora Kofun to an audience of about 80 people at the 1999 conference of the Astronomical Society of Japan on October 10, 1999. Dr. Miyajima is Associate Professor of History and Philosophy of Science at Doushisha University in Kyoto. When archaeologists found the painting of stars on the ceiling of the tomb, Professor Miyajima was asked to join the research group to study their astronomical significance. (Photo by Steve Renshaw)

Kitora tomb itself is approximately 1 meter in width, 1.3 meters in height, and about 2.2 meters in length. The painting on the ceiling is remarkably small yet detailed. Professor Miyajima developed a chart based on his plotting of stars, connecting lines, and circles found on the ceiling painting (see below). The outer or horizon circle is about 63 centimeters in diameter. The celestial equator and ecliptic circles are both from 41 to 43 centimeters in diameter, and the inner circle (denoting stars which were seen to not set in the North) is about 18 centimeters in diameter. The ceiling of the tomb has a slight slant at the edges. On the east wall, where the "blue dragon" was found, a painting of the sun was found at the ceiling's edge. On the west wall, where the "white tiger" was painted, a representation of the moon was found in a complementary position. Unlike Takamatsu Zuka Kofun, no paintings of human figures were found. Additionally, lines which connect stars in the Kitora painting appear to have been engraved and filled with a vermilion substance. While in Takamatsu, gold leaf was used to denote stars, stars in Kitora Kofun were carved with no filling.

In his presentation, Miyajima, who has published extensively on the subject of astronomical charting in Asia, made several points regarding interpretation of paintings of charts such as that found in Kitora Kofun. Generally, patterns of stars are symbolic. Charts such as those used as a reference in the painting of Kitora's ceiling were used for ritual and divination purposes. Images such as those found in many Western tombs are almost always absent, and delineations of stars and connecting lines usually serve a more decorative function in tomb representations than astronomical accuracy. This seems certainly the case with both Takamatsu and Kitora Kofuns, although stars in the latter are drawn more realistically and with more accuracy than those in the former. Observationally, for most ancient charts, only certain stars (almost always key stars for moon stations) were charted in correct positions. [Readers may want to refer to Needham (1959), Nakayama (1969), Nivison (1989), and Watanabe (1987) as well as review our introduction to star charts and moon stations.]

There are circular and square star charts. Takamatsu Zuka Kofun is obviously square while the chart in Kitora Kofun is round. Virtually all star charts based on Chinese practices are centered on the North Pole, and this is certainly the case with Kitora's painting. As in all Ancient Asian star charts, stars are not differentiated according to magnitude. Of charts that use circles, the oldest extant manuscript of such is one from China dated 1247 A.D. Miyajima noted in his lecture that the painting in Kitora Kofun does not match any extant star chart in China, Korea, or Japan. While there were, of course, many star charts older than that used for Kitora, none have been recovered so far, and the painting in Kitora Kofun could arguably be considered the oldest extant astronomical star chart. That is, it is a chart which not only serves a decorative purpose but also has designations of horizon, equatorial, and ecliptic circles as well as recognizable patterns based on dots representing actual stars. Many ancient representations in the West (such as Dendera in Egypt; see Gingerich, 1992) contain pictures of recognizable constellations but do not generally have actual patterns of discernable stars.

While Miyajima is somewhat forceful in his argument that the Kitora chart is the oldest extant "astronomical" chart, it should be noted that it is a painting designed to give the tomb's occupant a "place" in the cosmos. It was not designed to be used for even practical divination, much less observation, and the actual chart on which it was based has not been recovered. Further, as is discussed below, the chart does have obvious graphical and observational errors, at least to some degree attributable to a lack of understanding on the part of the artist of the underlying astronomical principles. It is difficult to determine whether the Kitora chart was based on Korean or Chinese models. Charts in Korea often had different sized stars (unrelated to magnitude) while those in China seldom if ever contained stars of different size. Given the above qualifications, the chart represented is remarkable in its delineation of the sky.

Composite Image of the Star Chart on the Ceiling of Kitora Kofun. (From Research Report of Cultural Heritage in Asuka Village Vol. 3; Scientific Research on Kitora Tumulus)

Accuracy of the Kitora Astronomical Star Chart

As noted above, ancient East Asian star charts, whether used for "professional" purposes or other, all have singular representations of stars in each constellation. Many of these constellations are unrecognizable to Western eyes (we have indicated a few which may be discerned in the chart below). Usually, only one star in a constellation (especially those of moon stations) had observational accuracy. Other stars in the pattern might be identifiable but were drawn relative to symbolic, political, or other purposes. In the Kitora chart, it is obvious that star groupings are not correctly spaced or sized. As an example, readers may note how big the Pleiades (18th moon station) appear in the chart below. Connected with criss-crossed lines, only one star (17 Tau), the key star for the Pleiades (see moon stations related to the "white tiger") has probable observational accuracy. From Miyajima's analysis, key stars along with certain stars in the Northern area of the Kitora chart appear to be observationally accurate.

[Large Kitora Chart]
From chart developed by Professor Kazuhiko Miyajima which matches details found on the ceiling painting. We have added lines and notations for reference.

One significant and as yet unexplained anomalie in the Kitora chart is the fact that the autumnal and spring equinoxes are clearly in error, even accounting for precession into any epoch. However, the error does not seem to be one of only careless drawing. There are several ways to view this error. Note that the chart equinoxes are both equidistant from a line which would intersect the north pole of the chart and the celestial equator at opposite points (refer to the chart above where we have drawn green lines from chart equinoctial points to opposites on the celestial sphere, and have bisected the consequent angle with a blue North/South "chart" line). Equinoxes should, of course, be at opposite sides of the celestial equator. Miyajima also notes that these equinoctial points could be symmetrical with a North/South line. We have illustrated in the chart above (note yellow lines) where if the chart equinoxes are mirrored east to west along the tomb's North/South line, a perhaps more accurate position of the equinoxes would be established. Finally, we did a simulation of the sky at 65 BC (see dating procedures below) using The Sky by Software Bisque and our own delineation of key star positions. We have also noted these in the chart above. It was common practice for ancient cartographers who drew circular charts to draw the ecliptic as a circle, a practice which continued well into the Edo Era (1603-1867; See Watanabe, 1987). Of course, because the ecliptic is at an angle to the celestial equator, a circle on a planar chart could never accurately portray the oblique shape of the ecliptic of which any point would be in spherical angular measurement from the celestial equator. However, in most ancient charts, at least the points of the equinoxes were generally drawn with accuracy, something that the artist of Kitora Kofun obviously did not do. It should be noted that Chinese and later Korean and Japanese chart makers did realize these problems and began to use "square" charts or what we would term more Mercator projections. On such charts, the ecliptic could be drawn with more accuracy if one wished to do so (See Needham, 1959 and again Watanabe, 1987; readers may also wish to refer to Yasui's 1699 chart which we have produced in our article on star charts.

[Miyajima 2]
Professor Miyajima explaining the location of Orion and pointing out anomalies relative to the equinoxes on the Kitora Chart. (Photo by Steve Renshaw)

Latitude of "Observer" and "Date" of the Chart

In most circular East Asian star charts, a Seikyo Houi Zuhou graphical rule is used. Basically, this rule is that the planar distance from the north pole (center as indicated on the chart) to any star (or other point on the chart) is proportional to the spherical angular measurement from the celestial north pole to that star (called Kyo Kyoku Do in Japanese). Further, of course, this ratio should be constant for all stars. In the chart above. the small inner circle marks those stars which are seen to not set in the North; hence the distance from the center of the chart to this circle (the circle's radius) should be proportional with the angle from the north pole to the observer's northern horizon which, of course, would be the observer's latitude. The distance from the center of the chart to the celestial equator should be proportional with 90 degrees [Ancient cartographers using the Kyo Kyoku Do would consider the celestial equator to have a celestial latitudinal measure of 90 degrees rather than 0 degrees declination as in modern usage; stars south of the celestial equator would have a Kyo Kyoku Do greater than 90 degrees]. The outer circle of the chart above represents the horizon of the observer. Stars outside this circle could not be seen from the observer's latitude. Using the principles outlined above, the reader may see that the distance represented by a line drawn from any point on the inner circle through the center of the chart to a point opposite on the horizon circle should be proportional to the observer's horizon to horizon angular measure. The chart below may help in visualizing these measures:

[Chart Measures]
Measures on Circular Asian Star Charts. (Graphic by Steve Renshaw)

Based on the above methodology, the ratio of the radius of the inner circle of the star chart to the observer's latitude should be equal to the proportion of the radius of the celestial equator circle to 90 degrees. Using the chart supplied in the Research Report of Cultural Heritage in Asuka Village Vol. 3; Scientific Research on Kitora Tumulus, we measured the radius of the inner circle at ~19.5 mm and the radius of the celestial equator circle as ~46 mm. The following equation should hold: 19.5/L = 46/90 where L is the observer's latitude. Solving for L, we got a value of 38.15. Miyajima indicates a value between 38 and 39. Of course, any value cannot be considered absolute as errors may come from at least three sources: plotting error by the artist, distortion due to the optical probe and consequent computer graphics image used, and just plain uncertainty with regard to original observation. Miyajima points out that measures for ancient star charts were generally made based on observations from the ancient capitals. The Nara area of Japan has a latitude around 34 degrees. Hence, Miyajima looked for an ancient capital in Japan, Korea, or China which would have had a latitude approximating 38 to 39 degrees. The closest would be the capital of ancient Kokuryo (close to present day Pyong Yang). That there was interaction between Japan and this kingdom in the Asuka era is well documented [See, for example, Sansom (1974) and Kidder (1993)], and it seems most probable that the chart used in the Kitora painting came from there. However, the manner of plotting stars seems to have originated in China. Like the Shih Shen star manual (Han Dynasty), the chart in Kitora Kofun utilizes observational data for plotting star coordinates that would have been in use in China around 70 BC (late Ch'ien-Han Dynasty) (See Needham 1959). This system was probably used by an astronomer in Kokuryo who adapted it for that location's latitude. However, such adaptation was apparently not made when the chart was brought to Japan.

The most difficult task for Miyajima and associates seems to have been the determination of the date of the chart. To accurately determine a probable interval, he had to determine where on the chart was "the center" or north pole and whether or not it matched the point indicated by the artist. Along with this, he had to determine where Spring Equinox would actually be based on positions of stars. To accomplish this task, he started with a theoretical base that may be best understood as follows: First, we know that for any epoch, every star has a Kyo Kyoku Do (easily determined from the star's declination) and a celestial longitude or right ascension. Second, we also have a chart sitting in front of us whose key stars probably have a certain degree of accuracy but may suffer from the same error sources mentioned above. The question would be: Is there an epoch where the differences between the ratios of planar chart distances to declination for all "reliable" stars are a minimum? If we can find such a "time point", then from the celestial longitude or right ascension we have established, we have a theoretical Spring Equinox and some time boundary for a consequent "date of observation".

Miyajima used a least squares method in which he sought to determine that point in time where selected stars theoretically believed to be accurately plotted would have right ascension and declination proportional to planar chart measures to a central point (north pole), differences in said proportions to be a minimum.

[Star Least Squares]
Each line from a star (denoted S1, S2, ...) to a central point on the chart (north pole) should be proportional to celestial angular measurement from the north pole to that star. If positions of stars are matched such that differences in these proportions are a minimum across all stars, then the central point should be a "best estimate" of the center or north pole of the chart.

Utilizing this method and perhaps being a bit overly cautious (or ambiguous), Miyajima came up with a date of 65 BC "plus or minus a few hundred years" as he put it. In the Asuka research report, his estimations are based, of course, on a curve whose "minima" is at -65 with plus or minus 250 years to "out-of-the-ball-park". True "North" was found to be only slightly different from the chart position. His line of best fit may be seen in the chart below:

[Least Squares Graph]
Least Squares "line of best fit" for data points calculated for 65 B.C. Note that the slope is less than 1, and there is obvious "scatter". Such are probably due to error sources mentioned in the text. (From Miyajima, 1999a)

As an exercise for readers with planetarium software, a star in Libra may be used as an example. The chart of star positions used earlier (above) has Libra marked, and the star which seems to rest just "a little north" of the celestial equator is an "accurate" star for the third moon station, TomoBoshi or Ti in Chinese. If your software allows for precession and changes equatorial grids accordingly, you may want to see how Beta Libra lines up with the celestial equator both before and after 65 BC.

Considering earlier comments on the probable date of Kitora Kofun's construction, readers should also note that the date of observation reflected in the star chart probably has nothing in common with the date of the tomb's construction.

Prospects for Research

Miyajima's work with Kitora Kofun concerns primarily the astronomical facets of the star chart, and from the point of view of History of Astronomy, provides a fascinating look into ancient chart making. Much of Professor Miyajima's work is, of course, "hoping for the best" when it comes to the accuracy with which the chart was created. However, his theoretical framework provides a good base of hypotheses which will hopefully see the light of interdisciplinary research. Many scholars have contributed to work on Kitora Kofun; we have listed only the principals in the reference (below) to the Research Report. Unfortunately, while many scholars have contributed, little seems to have been shared in terms of interdisciplinary discussion. While there is not an outright antagonism between fields such as astronomy and archaeology, many of the riffs found in the West are magnified in Japan, and a sense of respect (perhaps more than disdain) seems to keep scholars within the bounds of their own disciplines. From the point of view of archaeoastronomy and astronomy in culture, the door remains quite open to research in questions related to the socio-cultural milieu in which the tomb was constructed, the interaction of Chinese and indigenous Shinto cosmologies at work in Asuka times, and the ways in which such cosmologies played a part in the lives (and deaths) of Asuka citizens. In the future, we hope to contribute to such interdisciplinary study in this intriguing area of Japan's Astronomical History.


Gingerich, O. (1992) The Great Copernicus Chase and Other Adventures in Astronomical History. Sky Publishing Corporation and Cambridge University Press.

Inokuma, K., K. Izumimore, K. Kawakami, M. Sawada et al (1999) Research Report of Cultural Heritage in Asuka Village Vol. 3; Scientific Research on Kitora Tumulus. Asuka Village Board of Education. Asuka, Nara, Japan.

Kidder, J.E. (1993) The Earliest Societies in Japan. In Brown, D.M. (Ed.) The Cambridge History of Japan, Vol. 1, Ancient Japan. Cambridge University Press, Cambridge, England.

Krupp, E. C. (1983) Echoes of the Ancient Skies: Astronomy of Lost Civilizations. Oxford University Press, Oxford.

Nakayama, S. (1969) A History of Japanese Astronomy; Chinese Background and Western Impact. Harvard University Press, Cambridge, Mass.

Miyajima, K. (1999a) Kitora Tumulus Astronomical Chart. In Research Report of Cultural Heritage in Asuka Village Vol. 3; Scientific Research on Kitora Tumulus. pp. 51-63.

Miyajima, K. (1999b) Kofun no Naka no Uchuu - Kitora Tenmon Zu to Chuugoku no Uchuukan (Universe in Tumulus -Kitora Astronomical Chart and Chinese Cosmology). Presentation at the Conference of the Astronomical Society of Japan, Fukuoka, Japan, 1999b.

Needham, J. (1959) Science and Civilization in China; Volume III, Mathematics and the Sciences of the Heavens and the Earth. Cambridge University Press.

Nivison, D.S. (1989) The origin of the Chinese lunar lodge system. In Aveni, A.F. (Ed.) World Archaeoastronomy. Press Syndicate of the University of Cambridge, New York, pp. 203-18.

Sansom, G.B. (1974) A History of Japan to 1334. (Tuttle Edition). Charles E. Tuttle Co., Tokyo.

Sugaya, F. (1994) Nihon no Kodai Iseki (Ancient Japanese Ruins) Vol. 7, Nara Asuka. Hoikusha, Osaka.

Terao, I. (1996) Asuka Rekishi Sanpo (Asuka Historical Promenade). Sougen Sha, Osaka.

Uno, S. et al (1991) Nihon Zenshi (Japan Chronik). Kodansha, Tokyo.

Watanabe, T. (1987) Kinsei Nihon Tenmon Gakushi; Vol II: Kansatsu Gijutsu Shi. (A View of Japanese History of Astronomy in Modern Times; Vol II: Techniques of Observation). Koseisha Kosaikaku, Tokyo.

Please send comments to

Steven L. Renshaw

Return to Astronomy in Japan Home Page