Monograph No. 48

Reading the Past in Tree Rings: Development of Dendrochronology in Japan

Contents

 

Preface

Explanatory notes

I. Dendrochronology and its Previous Researches  3

by TANAKA Migaku

A. Out line of Dendrochronology  3

B. Previous Researches    7

1. Beginnings of dendrochronology and progress in Europe and North America  7

2. History of dendrochronology in Japan  11

 

II. Samples for Dendrochronological Research and its Methodology  18

 by MITSUTANI Takumi

A. Samples and Measurement of Width  18

1. Types of samples  18

2. Sample acquisition and preparation  19

3. Measurement s of tree ring width  20

B. Cross- Correlation of Tree- Ring Patterns  21

1. Drawing tree- ring pattern graphs  21

2. Cross- correlation of tree- ring patterns by numerical data   23

3. Recognition of key signature patterns  27

 

III. Examination of Ring Patterns utilizing Living Trees  29

by MITSUTANI Takumi

A. Examination of Features of Ring Patterns of Living Hinoki Cypress  29

1. Acquisition of hinoki cypress samples  29

2. Ring patterns of the same tree taken radially in different directions at the same level  31

3. Ring patterns at the center and near the exterior of the same tree at the same level  34

4. Ring patterns taken at different height of the same tree  34

5. Ring patterns taken from different trees in the same region  38

6. Ring patterns taken from trees of different region  45

B. Ring Patterns of Trees of Non-Cypress Species  49

1. Ring patterns of hiba arborvitae and sugi cedar  50

2. Ring patterns of koyamaki pine and other species  55

 

IV. Building Index Master Chronology  60

by MITSUTANI Takumi

A. Building Standard Chronology of Hinoki Cypress  60

1. Building standard chronology of modern trees  60

2. Confirmation and extensions of standard chronology  67

a. Standard chronology derived from structural parts of the Prayers Hall of the Nigatsu-do 二月堂 at Todai-ji 東大寺, Nara prefecture  68

b. Standard chronology derived from artifacts discovered at the site of the Kiyosu 清洲 Castle Town, Aichi Prefecture  71

c. Standard chronology derived from artifacts discovered at the Kusado Sengen 草戸千軒 site, Hiroshima Prefecture  73

d. Standard chronology derived from artifacts discovered at the Toba 鳥羽 Detached Palace site, Kyoto prefecture  74

e. Standard chronology derived from artifacts discovered in the Heijo 平城 Capital, Nara Prefecture  76

f. Extension of standard chronology beyond the Time of Christ   78

B. Standard Chronology Derived from Non-Cypess species  79

1. Comparison between ring patterns derived from different species of trees in the same region  79

2. Comparison between ring patterns derived from different species of trees in different regions  80

3. Standard chronology of sugi cedar  82

a. Standard chronology derived from living sugi cedar   82

b. Standard chronology derived from archaeologically discovered artifacts and structural parts in the Tohoku area   83

c. Extension of the standard chronology of sugi in the Tohoku area   86

d. Mean ring pattern of securely dated wooden artifacts discovered in Shizuoka Prefecture   87

4. Standard chronology or koyamaki pine  89

a. Standard chronology derived from artifacts discovered in Nara Prefecture   89

b. Floating chronologies  92

 

V. Application of Index Master Chronology   94

by MITSUTANI Takumi and TANAKA Migaku

A. Introduction   94

B. Application of Dendrochronology  98

1. Dendrochronological dates of living trees  98

2. Dendrochranological Dates of Artifacts 1  99

a. Frames of a well: Ochiai 落合Ⅲ site, Iwate Prefecture   99

b. Posts: Hotta no Saku 払田柵 fort site*, Akita Prefecture   99

c. Structural parts of houses buried at the Kurumidate 胡桃館 site, Akita Prefecture   100

d. Posts: Kinowa no Saku 城輪柵 fort site*, Yamagata Prefecture   101

e. Coffin: Nanamawari-Kagamizuka Kofun 七廻り鏡塚古墳 burial site, Tochigi Prefecture   101

f . Funerary urn: site of the Castle Town of the Asakura 朝倉 Family of Ichijodani 一乗谷**, Fukui Prefecture   102

g. Wooden objects: Yamagi 山木 site, Shizuoka Prefecture   102

h. Preserved pillars: Shida Gunga 志太郡衙 county- office site*, Shizuoka Prefecture    103

i. Wooden objects: Kiyosu 清洲 Castle Town site. Aichi Prefecture   104

j. Structural parts: a bridge attributed the Seta no Karahashi 瀬田唐橋, Shiga Prefecture  104

k. Preserved pillars: Miyamachi 宮町 site, Shiga Prefecture   104

l. Coffin: Jodoji Kofun 浄土寺古墳 burial site, Kyoto Prefecture  106

m. Coffin: Kawaradani 瓦谷 site, Kyoto Prefecture   107

n. Parasol-shaped object: Moat of the Kondayama Kofun 誉田山古墳 burial site (attributed to the Mausoleum of Emperor Ojin 応神), Osaka Prefecture  107

0. Coffin: Kyozuka Kofun 経塚古墳 burial site, Osaka Prefecture  107

p. Frames of a well: Ori’ono 遠里小野 site, Osaka Prefecture   108

q. Parasol-shaped objects: Shijo Kofun 四条古墳 burial site, Nara Prefecture   108

r. Parasol-shaped objects: Obaka Kofun 小墓古墳 burial site, Nara Prefecture   109

s. Part of a water clock: Mizu’ochi 水落 she*, Nara Prefecture  109

t. Preserved lower portions of pillars: Lower stratum of the Hokke-ji 法華寺 temple site, Nara Prefecture   110

u. Votive tablet of a horse: Higashi-Ni Bo 東二坊 [ East Second Ward] on the Nijo Oji 二条大路 [Second Major Street], Heijo 平城 Capital, Nara Prefecture  111

v. Frames of well: 13th and 14th Tsubo 坪 [Blocks], Ichi Bo 一坊 [First Ward] on Hachijo Oji 八条大路 [Eighth Major Street], Ukyo 右京 [Western Sector] , Heijo Capital  112

w. Trough: location of the historically known Masuda no Ike 益田池 pond, Nara Prefecture   112

x. Artifacts discovered at the Kusado Sengen 草戸千軒 site, Hiroshima Prefecture   112

y. Bases of a circular wooden-bent boxes: Shimokawazu 下川津 site, Kagawa Prefecture   113

z. Lower preserved portion of a pillar: Ono 大野 Castle site**, Fukuoka Prefecture  114

3. Dendrochronological dates of architectural structures   114

a. Boards used for wall: Kagura-den 神楽殿 [Hal l of Ritual Music and Dance]*, Ha' ushiwake 波宇志別 Shrine, Akita Prefecture   114

b. Pillar: Main Hall*, Wakamiya Hachiman 若宮八幡 Shrine, Nagano Prefecture  115

c. Structural parts: Three story pagoda*, Kiyomizu-dera 清水寺 temple, Kyoto Prefecture  116

d. Structural parts: Main Hall*, Gaya-in 伽耶院 temple, Hyogo Prefecture  116

e. Central pillar: Five story pagoda**, Horyu-ji 法隆寺 temple, Nara Prefecture  116

f. Central pillar: Three story pagoda*, Hokki-ji 法起寺 temple, Nara Prefecture  118

g. Structural parts: Otataneko 大直禰子 Shrine*, a subordinate shrine of the Omiwa 大神 Shrine, Nara Prefecture  118

h. Structural parts: Main Hall*, Hato-ji 宝憧寺 temple, Nara Prefecture   118

4. Dendrochronological dates of art objects  120

a. Lacquered wooden bent-box kepi at a temple in Kyoto Prefecture   120

b. Seat of the Bhaisjyaguruvaiduryaprabha statue** in the Image Hall, Horyu-ji, Nara Prefecture   120

c. Hyakuman-to 百万塔 [one million miniature pagodas] at Horyu-ji, Nara Prefecture   120

d. Guardian figure (the one in a pair who keeps the mouth colsed) in the South Gate, Todai-ji 東大寺 temple, Nara Prefecture   122

e. Seated Amitabha at Ganki-ji 岩崎寺 temple, Yamaguchi Prefecture   122

f. Seated Amitabha at Hoko-ji 法光寺 temple, Yamaguchi Prefecture   124

g. Standing Arya-avalokitesvara at Hoko-ji temple, Yamaguchi Prefecture   124

h. Standing Vaisravana at Hoko-ji temple, Yamaguchi Prefecture  124

i. Standing Acalanatha at Hako-ji temple. Yamaguchi Prefecture  125

j. Standing Virupaksa at Gatsurin-ji 月輪寺 temple, Yamaguchi Prefecture   125

5. Dendrochronological dates of buried forests  126

a. Sugi cedar in the Yamamoto Country, Akita Prefecture   126

b. Sugi cedar in the Yuri County, Akita Prefecture   126

c. Sugi cedar in the Mogami Country, Yamagata Prefecture   127

d. Hinoki cypress in Susono City, Shizuoka Prefecture   127

 

VI. Dendroclimatology and its Applications   128

  by SATO Tadanobu and YAJIMA Atsushi

A. Detecting Characteristic Long-Term Climatic Fluctuations Based on Changes in the Tree-Ring width   128

1. Reconstruction of palaeo-climate and the examination of data  128

2. Standardization of tree-ring width data  130

a. Method of standardization   130

b. Tree-ring width data of living trees   131

3. Climatic data  133

B. Detecting Characteristic Climatic Fluctuations  140

1. Correlations between tree-ring width and climatic factors  140

2. Estimation of climatic data utilizing correlation functions  144

a. Overview of the methods of estimation  144

b. Results of the estimation   145

3. Estimation of climatic data by the Kalman filtering algorithm  147

C. Conclusions   157

 

VII. Development of Dendrochronology in Japan   159

  by TANAKA Migaku

References  163

English summary  169

 

1 In this section,* and** respectively denote a “Historic Site” and a “Historic Site of Special Significance” designated by the national government .

2 In sections 3 and 4,* indicates an “Important Cultural Property,” **a “National Treasure.”

 

 Dendrochronology in Japan

 English Summary

 This report summarizes a research into dendrochronology and dendrochronology conducted principally by archaeologist TANAKA Migaku 田中琢 and archeo-botanist MITSUTANI Takumi 光谷拓実 at the Nara National Cultural Properties Research Institute.  The principal investigators are joined by SATO Tadanobu 佐藤忠信 of Disaster Prevention Research Institute at Kyoto University in the application of dendroclimatology. Our five year project initiated in 1985 has been supported by the Grant-in-Aid for Scientific Research program of the Japanese Ministry of Education. Prior to the presentation of our results, we first review previous research into the use of tree rings in Japan. We will proceed to discuss our studies, first on dendrochronology, and then on dendroclimatology.

 

 Studies based on tree rings in Japan originally began with the aim of reconstructing palaeoclimate and the cycle of palaeoclimatic fluctuations. These measurements were calculated on the basis of the width of tree rings. However, such researches always utilized only one tree; the scholars did not attempt to cross- date more than one sample. This was the case in all pre-World War II studies.

 The earliest example of dendroclimatological scholarship in Japan was HIRANO Ressuke’s 平野烈介 research published in 1921. He calculated the total amount of growth of a sugi cedar based on the area of the cross-section of the tree. He observed fluctuations in the amount of yearly growth and determined that: “In the life of this particular Japanese cedar tree, we have clear evidence of a thirty-three year cycle. “He identified this cycle as Bruckner’s cycle in climatology. Such studies were continued intermittently and other hypothetical periodic cycles of long-term climatic fluctuation were proposed - 110 years, 350 years, and 700 years. The most famous research was by YAMAZAWA Kingoro 山沢金五郎, then the director of the Weather Station at Takayama, Gifu Prefecture. In 1930, the published the result of his measurements of the width of 802 rings of a Japanese cypress tree (Chamaecyparis obtusa Endl.) and he attributed these 802 rings to the years between 1119 to 1920. In our re-examination, however, we discovered that these rings were formed between 1118 and 1919.

 Among the pre-World War II scholars interested in dendrochronology, architectural historian SEKINO Masaru 関野克 was an exception in that he not only followed the progress in this field in Europe and North America, especially the achievements of A. E.     Douglass who could indeed cross-date some floating chronologies, but also attempted to apply Douglass’ work to Japanese trees. In 1943, Sekino measured the width of rings of structural parts of an eighth century building in Japan, and same of the ring patterns he drew still survive. Probably because or the social confusion and disturbances during and soon after World War II, however, Sekino did not make any further progress in his research.

 Another epochal study was published by SHIDEI Tsunahide 四手井綱英 and his collaborators in 1944. Their project, conducted in Akita Prefecture, was methodologically significant because they utilized measurements from several hundred Japanese cedars rather than just one tree. They concluded that the fluctuation in the tree growth between 1928 and 1942, as evidenced in the different width of tree rings, was related to the precipitation coefficient 1, the growth of Japanese cedar became poor whenever the precipitation coefficient went above or below certain values.

 The majority of post war tree ring studies in Japan were climatological. Scholars became methodologically precise and were longer satisfied with simple comparisons between ring patterns and climatic fluctuations. Nonetheless, meticulous data-gathering such as characterized Shidei’s work was not yet common.

 Two studies conducted during that period were worthy of note. One was TAKAHASHI Hiroaki’s 高橋宏明 study. Following Shidei’s example, Takahashi explored the relationship between tree growth and climatic factors, especially precipitation coefficient. He showed the correlation among the average width of rings of several trees at one specific site and the average width taken at different sites. The fact that trees of different ages and at different localities showed the same growth pattern indicated the connection with a common environmental factor - climate. This was the starting point of his research. Unfortunately, no scholars continued this line of research and thinking.

Another interesting study was conducted by NISHIOKA Hideo 西岡秀雄, who set out to date an architectural structure using dendrochronology. He measured the width of approximately 250 rings of the central pillar of the pagoda at Horyu-ji 法隆寺, Nara Prefecture, the oldest surviving wooden structure in the world. He then compared the measurements with a ring pattern taken from part of a different structure built in the 730s also at Horyu-ji and concluded that the tree used in making the Horyu-ji pagoda pillar was cut down before 607. Nishioka’s work was significant because it was t he first use of tree-rings for the purpose of dating. Nonetheless, both tree samples from which he took ring patterns lacked the outer most part of the bark, which would have given precise dates of trees cut down. Consequently, Nishioka’s work has been regarded by later scholars as hasty and imprecise and has been duly ignored.

Part of Japanese negative attitude toward dendrochronology came from scholars’ belief in the impossibility of its application to Japanese trees. They believed without any testing that the climate and topography in Japan were so different from Arizona where Douglass developed dendrochronology that what was possible in Arizona would not be possible in Japan.

What tempted us to pursue dendrochronology at the Nara National Cultural Properties Research Institute was that we had excavated numerous wooden artifacts at archaeological sites. We had also conducted ethnographical / historical field research into a number of ancient temples, shrines, and other old wooden structures. Our knowledge about European and American achievements in this field also stimulated our research.

In 1970, one of our colleagues started to measure tree ring width. After be processed some samples, he reached the conclusion that it would be very difficult to apply dendrochronological methods to Japanese cypresses. Since most buildings in Japan were traditionally made from cypresses, the project was halted.

In 1979 we started a preliminary research and found the application of deadrochronology and dendroclimatology to Japanese trees possible. In 1985, our major studies in these two fields began, and we confirmed the possibility of the application of dendrochronology to archaeology, history, architectural history, and art history. We also made considerable progress in dendroclimatology. At the same time of our research, other dendrochronological studies were published. Our studies could be distinguished by the fact of having established a significantly long-term ring pattern.

 

 The method for cross-correlation of tree-ring patterns that we used was basically same as that of Europe. In order to standardize the varying tree ring width, we used the five running mean method. To determine statistical significance, we used Student’s “t” test. With sixty or more degrees of freedom, we compared the obtained t value with the 0.1 per cent significance level of t, which was 3.5. Correlations of ring patterns which grew over the same span of years normally produced t values greater than t=3.5. Our experience showed that samples in a pair which yielded a t value greater than 3.5 did not necessarily cross-correlate. We always double-checked with the results of visual matching of pattern graphs with reference to key signature patterns. The statistical calculations were simply means of cross-correlating ring patterns.

 The materials were taken from living forest trees, old structures, and wooden artifacts discovered at sites. The kinds of trees used in our studies included: a hinoki cypress Chamaecyparis obtuse Endl., a samara cypress Chamaecyparis pisifera Endl., an asunaro arborvitae Thujopsis dolabrata Sieb. et Zuocc., a hiba arborvitae Thujopsis dolabrata Sieb. et Zucc. var. Hondai Makino, a kurobe arborvitae Thuja Standishii Carr., a tsuga hemlock Tsuga sieboldii Carr., a sugi cedar Cyptomeria japonica D. Don, and a koyamaki pine Sciadopitys verticillata Sieb. et Zucc. all of which were coniferous trees, as well as some deciduous trees, such as mizunara oak Quercus mongolica Fischer ex T. var. grosseserrata (B1.) Rehd. et Wils. and a buna beech Fagus crenata B1..

 The first step of our dendrochronological studies was to confirm the applicability of dendrochronology to Japanese trees and to determine appropriate species for our studies. For the first purpose, we selected a Japanese cypress because it was native to Japan, and natural forest of this tree was distributed from the southern half of the Honshu island to southern Kyushu. Cypresses had al so been most widely used for Japanese architecture, and were often discovered at archaeological sites. Utilizing modern Japanese cypress trees, we recognized the following features important for our studies: 1) ring patterns taken radially from the same tree tended to highly correlate with each other; 2) rings farmed during trees’ early life tended to show features unique to individual trees, this leading to the intelligence that for dendrochronological studies we should avoid rings near the center of a tree; 3) rings near or at the base of tree often showed anomalies because of the expansion of the root; 4) given a certain region, rings of different trees showed very high correlations to one another. We could even cross- correlate samples taken from trees approximately 400 kilo-meters apart.

 For the second goal, we tested a koyamaki pine, a sawara cypress, an asunaro arborvitae, a hiba arborvitae, a kurobe arborvitae, a tsuga hemlock, a sugi cedar, a mizunara oak, related species of which were widely used in European dendrochronological studies, and a buna beech which was the most representative deciduous trees in Japan. Among these, we found the first seven species suitable for our studies. It is ironic that the last two species were not found to be applicable.

 The next step was to build an index master chronology, which was the foundation of any dendrochronological studies. Our index master chronology of hinoki cypress trees spanned from 317, B.C. to 1984. It was derived from: Standard Chronology A (1695-1983) taken from a modern tree; Standard Chronology B (1027-1755) taken from a structural part of a sixteenth century building at Todai-ji 東大寺, Nara; Standard Chronology C (751-1591) derived from artifacts discovered in the Kiyosu 清洲 Castle

 Town site (fifteenth and sixteenth centuries), Aichi; Standard Chronology D (512-1322) derived from artifacts descovered at the Toba 鳥羽 Detached Palace (eleventh and twelfth centuries) site, Kyoto and those at the Kusado Sengen-cho 草戸千軒町 (Medieval Age) site, Hiroshima; Standard Chronology E (37, B.C.-A.D. 838) derived from artifacts discovered in the Heijo 平城 Capital (710-784) site, Nara; and Standard Chronology F (317, B.C.-A.D. 258) derived from artifacts discovered at the Yayoi (fourth century. B.C. to A.D. third century) and Kofun (fourth to sixth centuries) sites.

 We also built chronologies derived from sugi cedar trees, but none of them overlapped with one another. They were: Standard Chronology A (1779-1986) taken from   modern tree Standard Chronology B (405-1285) taken from artifacts discovered at sites in the Tohoku District and cross-correlated with the Index Master Chronology of cypress and Standard Chronology (420, B.C.-A.D. 365) derived from artifacts discovered at sites in the Tokai region (the Pacific Coast of the Chubu District) and cross- correlated with the Standard Chronology F of cypress.

 In addition, we built two Chronologies utilizing koyamaki pine trees. A 556 year ring pattern derived from artifacts discovered in the Heijo Capital cross- correlated with the Standard Chronology E of cypress and spanned from A. D. 186 to 741. In the same way, an artifact discovered at the Shijo Kofun 四条古墳 burial site in Nara gave rise to a chronology between 286 and 695.

 

 We then applied these chronologies to artifacts, early architecture, and crafts in an attempt to date them. Here are some examples: the Shijo Kofun, Nara, the Seta no Kara Hashi 瀬田唐橋 Bridge site in Shiga Prefecture, the Miyamachi 宮町 site, Shiga Prefecture, and the Hotta no Saku 払田柵 (Hotta Fort) site in the present Iwate Prefecture.

 At the Shijo Kofun, 448 was the latest dendrochronological date obtained from numerous artifacts made of koyamaki pine, which had presumably been used for mortuary rituals, and were discovered in the moat surrounding the Kofun. In other words, the Kofun was built sometime after 448. Since the absolute dating of kofun or burial mounds built between the late third and sixth centuries had always been debated, we believe that dendrochronology would be a useful method by which to determine absolute dates.

 We applied dendrochronology to try to estimate when a bridge was built for the first time over the Seta River. Archaeologists excavated structures of a bridge near the present Set a no Kara Hashi Bridge 3.5 meters below the bottom of the River. We utilized several long pieces of timber and lumber of 20 to 50 centimeters in diameter or in length. We confirmed that the bridge was constructed after 607, and consider it probable that this was the strategic bridge the control over which was an important aspect of a civil war in 672.

 A ring pattern taken from the lower portion of a wooden pillar discovered at the Miyamachi site cross-matched with the Standard Chronology E of cypress, and we could confirm that the tree for the pillar was cut down between 742 and the beginning of 744. This made it very highly likely that the site was a part of the Shigaraki 紫香楽 Palace which was occupied from the eight month of 742(on the lunar calender) and the fifth month of 745.

 In the same way, we could determine that the construction of the Hotta no Saku 払田柵 was begun between 801 and 802. There was a cross-matching of ring patterns yielded from longs of cedar (a square of approximately 30 centimeters in length) which surrounded the site and the Standard Chronology B of cypress. This refuted the hypothesis that the site was the historically known Okachi 雄勝 Castle built in 759.

 

 Our study of dendrochronology made considerable contributions to architectural history and art history as well. Examples of dendrochronological applications included Horyu-ji, a Nara Period horse painting, a Kamakura Period guardian figure at Todai-ji 東大寺, Buddha figures in Yamaguchi Prefecture, Ha’ushiwake 波宇志別 Shrine in Akita Prefecture, Wakamiya Hachiman 若宮八幡 Shrine in Nagano Prefecture and a Medieval wooden bent-box in Kyoto:

 We applied dendrochronology to date the western complex of the Horyu-ji. There had been two major hypotheses concerning the dating of Horyu-ji: a) the temple was built around 610 during the reign of Empress Suiko 推古;  and b) the temple was burnt down around 670 and rebuilt at the end of the seventh or the beginning of the eighth century. The number of rings of the central pillar of the pagoda in the western complex was 351, corresponding to the years 241 to 591. The pillar had been extensively modified from the original tree trunk by largely removing the exterior portion and the central portion; it would be unlikely that only the outermost twenty rings were removed for this modification. We found it very unlikely that the pagoda was built during the reign of Empress Suiko around 610.

 Next, we were able to give a more specific date to a small painting. In 198S we discovered in a garbage pit in the Heijo Capital a painting of a horse done on a cypress board of 27 by 19.5 centimeters and 0.7 centimeters in thickness. Since this board still kept the most exterior portion of the original tree, we could date the outer most ring to the year 728. We could also date the pit to 737 by wooden tablet with a dated inscription. The horse was probably painted between 728 and 737. This dating was felicitous because the horse was one of the few surviving examples of the eighth century paintings in Japan.

 Next, our ability to achieve this degree of specificity in dating, we could confirm a historical record. In 1181, one year after Todai-ji was burnt down in a civil war, priest Chogen 重源 and his collaborators began to reconstruct the temple. Two gigantic guardian figures (4.5 meters in height) , housed in the south gate of the temple, were built between the seventh month and eleventh month of 1203 (on the lunar calendar). In 1989, there was an overhaul of one of the figures, so we took the opportunity to apply dendrochronological dating to the pans of the figure which still retained bark. Original trees utilized for this statue were cut down in the winter of 1196, 1199, and 1201 or during the following spring of those years. Further, we discovered that one of the ring patterns taken from the statue was almost identical to that of a Buddha figure in the present Yamaguchi Prefecture - approximately 400 kilometers apart. Since it was historically known that Chogen brought timber for structural parts of the temple, it was quite likely that wood for the guardian figures also came from Yamaguchi.

 In other cases, finally, dendrochronological dates could contradict dates determined according to stylistic evolution. In and the vicinity of the Tokuchi 徳地 Town, Yamaguchi Prefecture, there were many Buddha figures in various temples, which were stylistically dated to the late twelfth century. We applied dendrochronology to fifteen of them and discovered that the most recent rings of trees used in six of the figures were formed in the 1190s. Since the exterior portion, and in some cases the central portion of trees, had largely been removed, the trees must have been cut down in the thirteenth century. The earlier dating should be reconsidered.

 Structural parts of the Kagura-den 神楽殿 (Hall of Ritual Music and Dance) at the Ha’ushiwake Shrine yielded dendrochronological dates of 1177 and 1195 by cross-matching with the Standard Chronology B of cedar. Since both of the structural parts kept the bark, the original trees were cut down around that time in the twelfth century. According to stylistic criteria, the Kagura-den had been estimated to date from the fifteenth century. When these Shrine structures are overhauled sometime in the future, we should more thoroughly investigate the dates of the structures by applying dendrochronology.

 Similarly, major pillars of the Hon-den 本殿 (Main Hall) at the Wakamiya Hachiman Shrine were dendrochronologically dated to be after 1614, the structure was stylistically considered to be of the Momoyama Period (1582-1600).

 Further, this method was effective in detecting forgeries. A certain temple in Kyoto Prefecture owned a wooden bent-box of cypress a cylindrical container) which had the inscription dated to 1233 on the bottom. We discovered, however, that the most recently formed tree-ring was dated to 1576. The container was forgery !

 In addition to architectural structures, art objects, and artifacts, dendrochronology is effective in dating natural disasters. A standing hinoki cypress of 1.4 meters in diameter with the bark completely preserved was discovered underground in the Susono 裾野 City at the foot of Mt. Fuji 富士.  Taking the topography of t he vicinity into consideration, it was probable that lava issuing from an eruption of Mt. Fuji dammed up a river, thereby submerging the whole forest and killing the trees. If this was indeed the case, we could estimate the date of an eruption of Mt. Fuji. By cross-matching with the Standard Chronology E of cypress, we could dale the standing tree to A.D.833.

 Similar discoveries of buried forests which might indiate natural disasters were reported from all over Japan. In the Mamurogawa  真室川 Town, Yamagata Prefecture,  some twenty cedar trees were similarly discovered six meters below the ground. By cross-matching with the Standard Chronology B of cedar , we confirmed that the most recent ring was formed in A.D. 850. In other words, a natural disaster which buried a forest hit this area between the winter of 850 and spring of 851.

 In Futatsu’i 二ツ井 Town, Akita Prefecture where the people believed in a tale that several thousand koku 2 of wild cedars were buried by a landslide during the Tokugawa Period (1600-1868), numerous large trunks of cedars and Japanese oaks of approximately 1.5 meters in diameter were discovered a few meters below the ground. We dated 754 rings of cedar tree by cross-matching with the Standard Chronology A of cedar and discovered that the most exterior ring was formed in A.D. 958. The forest was buried in the middle of the tenth century, much earlier than the people believed.

 In the southern part of Akita Prefecture, there was an area where a large forest was buried more than ten meters below the ground, Most of the trees were cedar and some exceeded f our meters in diameter. Taking into consideration the soil of the area and the way in which the trees had fallen, we hypothesized that an enormous mud flow resulted from an eruption of Mt. Chokai 鳥海 some ten kilometers southeast of the forest. Radiocarbon dates showed that the disaster happened abound 2600 B. P. We applied dendrochronology to eight somples taken from the buried forest, six of which were with barks completely preserved. The most recent rings were formed al the some time; it was likely that one eruption caused this complete destruction. The fact that the most recent rings contained fully mature cells indicated that the eruption occurred sometime between the winter of a certain year and the spring of the following year. We derived a ring pattern which spanned 848 years, but it remained as a floating chronology; it was too old to be cross-matched with any standard chronologies. Once we extend our Index Master Chronology to 2600 B. P., we could confirm the precise date of this disaster.

 

 In addition to the dendrochronological advances discussed above, we progressed in the application of dendroclimatology. The ring pattern we used was derived from thirty samples of living cypress trees in Nagano Prefecture and the Standard Chronology which was presumably based on timber produced in the same prefecture. We also collected the data concerning the number of rainy days from April to September in each year for the past 170 years from a local weather station and from diaries kept by local people before the weather station started. We filled missing data (years for which we did not have records f rainy days) by a one-dimensional autoregressive process. We discovered that our reconstructed 170 year-long data of rainy days per annum 3 closely corresponded to the current precipitation data in Nagano Prefecture. We concluded that the number of rainy days was appropriate for climatic data.

 In order to determine the most influential climatic factor in tree growth, we tried correlating the number of rainy days and average temperature to tree ring width. Tree grew fastest when there were many rainy days temperature did not rise.

 We then compared the power spectra of treering width and of the per annum rainy days. We discovered cycles of approximately six years, eight years, and twenty-three years to be common to both sets of data. This indicated to us a very good correlation between the tree ring width and per annum rainy days.

 Owing to this correlation, we proposed a system in which climatic data represented by per annum rainy days were input and tree ring width data were out put, by the process of an autoregressive moving average function. In order to identify per annum rainy days from the tree ring width data, we developed a correlation function. The most effective function for this purpose was a process with three autoregressive and two moving average coefficients. Using the identified parameters of this process, we reconstructed the per annum rainy day for eighty years beginning in 1760. We derived the relationship between the tree ring width data and per annum rainy days from 1009 to 1984, assuming the linear system we had used previous between the two factors. We standardized the primary tree ring width data for these 976 years by the spline function. We used the Kalman filtering algorithm, including the U-D observation updated theorem, to identify the system parameters. In the process of parameter identification, every parameter gradually converged to a certain value.

 By utilizing these system parameters, we calculated per annum rainy days for the 170 years for which we originally had good weather records. Our calculation was in basic agreement with the actual data. In the same way, we could reconstruct the per annum rainy days for approximately 800 years before 1813. These results of our research showed that the application of dendroclimatology was possible in Japan.

 

 Another major contribution of our research was the applicability of dendrochronology and dendroclimatology to Palaeolithic samples. At the Tomizawa 冨澤 site, Sendai City, Miyagi Prefecture, excavators discovered features three meters below the ground, which suggested that the inhabitants had been manufacturing stone tools at a camp fire. We could build a ring pattern spanning 323 years, derived from cross-correlation of three samples of genus Larix (a kind of larch). Since the site was dated by carbon 14 to have been more than 20,000 years old, we could not dendrochronologically determine the date of the site, as our Index Master Chronology did not extend beyond 317, B.C. Nonetheless, this result showed the potential application of tree ring studies to the Palaeolithic Period.

 

 In summary, our ten year project (including the preliminary research started in 1979) has shown that the application of dendrochronology and dendroclimatology to Japanese trees is possible. Our Index Master Chronology of cypress extends back to 317, B.C. The Master Chronology of cedar of the Tohoku District extends to A.D. 405, and that of cedar of the Pacific Coast of the Chubu District to 420, B.C. We have certainly progressed to the stage of application to other disciplines such as archaeology, architectural history, art history and the history of natural disasters in Japan.

 

1 The ratio between amount of annual precipitation and annual average temperature

 

2 a unit of volume used widely in Japan before the Meiji Restoration of 1898; in the case of trees, one koku equaled to 0.278 cubic meters.

 

3 Throughout this section by “per annum rainy days - we mean the total number of rainy days from April to September. Such data sets were chosen as the climatic data because tree rings grow most from spring to fall.

 

1990年8月31日 発行

年輪に歴史を読む

-日本における古年輪学の成立-

奈良国立文化財研究所学報(第48冊)

 

このページの先頭へ

上に戻る