National Information Service for Earthquake Engineering
University of California, Berkeley
Professor of Architecture
University of California, Berkeley
One of the surprises of the January 17, 1995, Hyogoken-Nanbu earthquake was the extent of the poor performance of traditional Japanese wood residences and rowhouse-stores. Although the construction system and design of these buildings differ greatly from structures in the United States, their inclusion in this report is important because their performance illustrates problems inherent in timber construction and poses crucial questions about whether people incorporate the observations of past seismic disasters in traditional building practice. These vernacular structures encode Japanese cultural and design values which are not being reincorporated in the two-by-four and prefabricated steel and wood structures which are replacing them.
Traditional Japanese buildings have been crucial to the development of modern architecture in the West. Gropius, Mes van der Rohe, and Wright have all been influenced by the spare beauty of these structures supported on elegant posts, whose plans, regulated by the module of the tatami mat incorporate varied symmetrical and asymmetrical spaces made even more flexible by removable shoji screens and shutters. The indoor-outdoor permeability of these elegant and seemingly ephemeral structures, the manner in which they combine interior design with exterior views of controlled landscaping and near vistas, made a deep impression on modern American architects.
Traditional residences are primarily of two types: the detached upper-class residences, and, less well know in the West, the traditional city rowhouses or machiya, which incorporate spaces for living and selling [Tingey, 1981; Durston, 1987; Shirnamura and Suzura, 1993]. These wooden structures have narrow street frontages of about 24 feet in which the store or restaurant is located, often entered through a toriniwa or stone walkway which leads both to the shop and back into the more private areas of the house which include rooms usually facing a rear garden. The construction system of the machiya is nearly identical to that of traditional residences. Many traditional machiya and updated Meiji (1867-1912) versions still stood in the Kobe area before, the 1995 earthquake.
Behind both types of structures at the end of the lot often stands a kura, a special storehouse to be used when earthquakes, fires, or hurricanes threaten [Itoh, 1974; Treib, 1976]. Built to be stronger than the Japanese house, the storehouse was intended to serve as a receptacle for the transfer of valuables when disasters were imminent. Several different types of kura can be found in Japan. In this area the kura is of the dozo type, a two story gabled storehouse with an interior framework of heavy timber coated with clay and covered with a fine plaster finish. The deep clay walls, tile roofs and shutters protected contents from fire. The heavy timber construction lashed or pieced together with complex joinery and encapsulated in deep clay walls could resist hurricanes as well. But these stout heavy structures could not resist seismic damage and failure.
A large number of traditional structures in downtown Kobe, and in Ashiya, Nishinomiya, and Nagata collapsed during the earthquake causing loss of propertyy and death. Traditional construction accounted for the vast majority of the fifty-five thousand wooden buildings that collapsed and the thirty-two thousand partially destroyed in the earthquake [AIJ, 1995]. Rowhouses, residences, and stores performed poorly, causing most of the five thousand earthquake casualties [AIJ, 1995]. The cause of this poor performance of traditional wooden construction lay in both the design of the structures and their deteriorated condition.
It appears that the aesthetic virtues of traditional Japanese structures praised by Western architects are the very features that undermine their structural integrity and make them vulnerable in earthquakes: delicate post and lintel construction, flexible plans and indoor-outdoor permeability created by screens and shutters. Many suffered major cracks in exterior walls and appeared out of plumb. All had some form of damage to their roofs. The heavy tile roof in which tiles are bedded in two or three inches of loose sand is designed to mitigate against fires and hurricanes. But it rests upon a structure which cannot resist the lateral forces such a weight generates. Further, if the roof is asymmetrical it often has little lateral integrity to work as a diaphragm in distributing forces.
The heavy roof rests on a structure which may itself be “light”, composed of 3 5/12 inch posts at 3 ½ or 6 or 9 or 12 foot intervals with mortise and tenon joinery secured by wedges and/or toe-nailing. While many joints held others did not have the strength either to resist, or the ductility to transmit, lateral forces. They either disassembled or shattered. The wooden post and lintel system, the “moment frame” of these structures, is not assisted by any shear walls and diagonal bracing is rarely present. When diagonal braces occur they are usually found at the corners of a building, attaching the corner posts to the floor plate. Commonly walls are constructed of a lath to which is applied a light armature of tied bamboo over which a fine mud plaster is applied. This kind of wall, while it looks beautiful, can develop no shear. The very few walls are useless in earthquakes.
The supporting posts often sit on a brick foundation or simply on the ground secured by the force of gravity alone. Traditional Japanese houses are often raised above the ground several feet to promote the flow of air under the ground floor. Continuous foundations are therefore rarely present. In the instances I saw there were no continuous foundations nor were there anchor bolts or brackets to secure the posts to the foundation. As at the roof level, the buildings were not tied together.
The earthquake appeared to have shaken the buildings back and forth until the weight of the roof acting on the joinery caused it to break and to settle out of plumb or to fall apart. In the worst cases the structure either collapsed straight down in a telescope fashion or more commorly fell to the side where it either hit a neighboring structure or dropped to the ground.
The stores in central Kobe, because of their narrow, open first floors could be said to have collapsed because of a soft story problem. But in truth one could say that all traditional Japanese construction suffers from potential soft stories because of its design characteristics. In the case of many buildings no diaphragms were incorporated in floors or roofs. There was no way to distribute the unequal torsional loads often generated by asymmetrical plans.
The dozo style kuras fared little better than the residences and machiya. The kura walls formed a mass which was too heavy for its wooden framework to stabilize. The narrow two story kuras rocked back and forth shedding their clay and plaster covering. Many stood out of plumb and I found several had collapsed entirely.
In several of the collapses surveyed it was obvious that the comections had rotted or had extreme termite damage. Whatever strength the structural system has depends upon its connections. Compromised by rot or insect infestation these structures have no capacity to resist earthquakes. A number of the rowhouses in downtown Kobe and in other areas were occupied by low-income tenants and owners, undoubtedly factors in why these structures were poorly maintained. However, as previously stated, the connections themselves were suspect, even without rot and insect infestation
The Takahashi/Kawai house (hereafter the Takahashi house) illustrates how a traditional Japanese house failed in the earthquake. Like many badly damaged traditional structures the house has since been demolished leaving photographs as the only records of how it fared.
The Takahashi house was located on alluvial soil one block east of the Miyagawa River between the Hanshin highway on the south and the Kokudo National Route 2 highway on the north in the Uchidekozuchi-cho neighborhood of eastern Ashiya. The wooden house had a garden on south and east sides. To the north, adjoining the wooden house, is a post-war reinforced concrete house also owned by the family which suffered little damage in the earthquake and still exists. To the west an apartment building, which also suffered no damage in the earthquake, hovers just beyond the property line. Between the Takahashi house and the apartment building is a concrete wall erected by the apartment house owners over the objections of the Takahashis.
The Takahashi house was an adaptation of the shoin style. Shoin literally means writing desk and first appeared during the Muromachi period in reference to study areas used by abbots in Zen monasteries. These study areas broke down the formality of the earlier shinden style halls. In the shoin style the proportions of the dwelling were based on the tatami mat, which covered the entire floor. The Takahashi house incorporates characteristic shoin style features: the decorative alcove (tokonoma) in the formal main room with its staggered shelves (chigaidana) and built in desk (tsukeshoin); the structure supported by square posts; and the coved, decorated ceiling of the formal room. Interior spaces can be divided by pulling shut plain or shuttered screens (fusuma) while the garden vista can be closed by shoji screens or heavy wooden panels (amando). The main room is flanked by two rooms to the north. There is a second story above the main room and a second story addition overlooking the garden to the east of the main room. On the northwest corner of the property is the family kura.
The construction system of the house was traditional post and bearn with unreinforced mud and bamboo walls. The beautifully crafted posts and beams were joined by mortise and tenons and were spaced to support maximum spans to provide the open areas so vital to the shoin style dwelling. At the time of the earthquake the joinery was in excellent condition showing no signs of insect infestation or dry root. Terra-cotta tiles typically found in the Kyoto-Osaka area were placed in a bed of sand on the roof
During the earthquake the building oscillated in the east-west direction. As it fell in an apparent final collapse to the west the entire back wall hit the contested concrete wall on the property line. The wall arrested the collapse saving the matriarch of the family sleeping on the first floor, but utterly shattering the walls of both the house and its kura. Mortise and tenons seem to have completely separated, wall joints toe nailed in place or secured with wedges uncoupled. The mud, sand and bamboo walls shattered. Every connection in the house was badly strained or broken, not a wall or floor remained plumb. The kura shook loose its mud covering laying bare the heavy timber construction.
Before condemning the traditional construction of the Takahashi house it is important to remember that there were high accelerations in this area: the box girders of several modem steel and concrete megastructures at Wakaba-cho just a kilometer south snapped apart. A block north the bottom floor of a reinforced concrete apartment building collapsed. Also within a kilometer to the east the Hanshin elevated highway collapsed. Yet across the street from the Takahashi home several steel and wood prefabricated residences survived unscathed.
After examining the poor performance of traditional wooden buildings we might ask why they were not constructed to be more seismically resistant. The building practices in the Kobe area are not so different from those throughout Japan. In this country where earthquakes are so common why has traditional architecture apparently failed to respond? The answer to this question is complex and multifaceted, cultural as well as technological. I present only a few preliminary observations here which I hope to investigate in more depth.
The literature on the seismic resistance of traditional wooden construction is scarce. Two representative commentaries on traditional Japanese architecture mention natural disasters like earthquakes in their discussions of climate: “Japanese architecture came into being under a feudal regime, in a damp climate; it makes use of perishable and replaceable materials, which have to be light in weight on account of the earthquakes” writes Andre Corboz [Corboz]. Heinrich Engel in his fundamental text, The Japanese House --- A Tradition for Contemporary Architecture, argues that since earthquakes “afflict man and his buildings in a manner similar to that of the seasonal storms, that are, in an architectural sense, of the same environmental nature.” Engel lists earthquakes along with typhoons, heavy precipitation, snowfall and humidity as problems [Engel, 19641.
Adaptation to climate makes traditional Japanese architecture very dangerous in earthquakes. The structures are often raised on stilts to avoid humidity. The interiors have as few walls as possible to provide maximum ventilation during insufferably hot summers. Shoji screens can be shifted or removed to provide visual privacy while causing little impediment to the circulation of air. In the winter the shoji screens and shutters can be remounted providing some insulation, but certainly not the kind of warmth westerners might expect. Heat is provided by portable charcoal braziers and kerosene and electric space heaters which were so deadly in starting fires during the Hyogoken-Nanbu (Kobe) earthquake.
Engel points out that Japanese residential construction is as delicate as it seems and not adapted to either typhoons or earthquakes: “The entire framework, owing to the lack of any diagonal bracing members, is susceptible to the slightest horizontal stress, the danger of which is aggravated by a fatally high point of gravity due to the overly heavy roof. Indeed, the Japanese house in a typhoon is like a house of cards in a draft.” Similarly Engel finds an absence of any measures to counteract the effects of earthquakes and doubts whether the Japanese even considered the elasticity of wood in relation to seismic resistance.
The lack of lateral stability in the framework certainly must raise doubts as to whether the Japanese chose wood material because they realized its advantages in the face of earthquakes. Rather, it seems that the Japanese actually were never really aware of the structural advantages of wood ... In spite of continuously recurring earthquake damage, the over-dimensioned roof construction and heavy roof load of clay ties and clay joinery have been maintained, without any visible attempt to reduce the dangerous top weight or to increase the dimensions of the under-sized vertical members, or to brace the latter by rigid diagonal members. Even minor horizontal earthquake shocks, thus, may easily become fatal...” [Engel, 1964]
Japanese authors also discuss earthquake resistance in traditional structures summarily and fatalistically. Teiji Itoh uses the example of the great Kyoto earthquake of 1830 to illustrate a Japanese attitude toward seismic disaster. So forceful was the earthquake that it collapsed the wood and mud storehouses (or kura) of the rich prompting the common people to write comic epitaphs on the broken walls ridiculing the structures and their owners. Itoh concludes:
This no doubt would not have happened if such destruction from earthquakes had been a common occurrence. It is precisely because the earthquake was of unprecedented violence that people felt moved to joke about the fallen storehouses. Perhaps we may see in this an example of the Japanese attitude toward earthquakes: They are unpredictable catastrophes that must be coped with by the human spirit, not by the structure of buildings [Itoh, 1974].
The extent of the incorporation of consciously anti-seismic features in traditional wooden buildings is an unresolved question. As early as the late 18th century the Jishin-den, or Jishingoten, the Earthquake Palace or Earthquake House, was constructed in the gardens of the Kyoto-Gosho, the Imperial Palace of Kyoto [Fujioka, 1956]. There is no doubt that whoever designed this structure wanted to make it earthquake resistant. It lacks the raised floor and heavy roof which are dangerous in earthquakes and incorporates a continuous foundation and oversized timbers for strength and continuity. During the Meiji period an active group of engineer’s and seismologists tried to design earthquake resistant buildings, among them John Milne and F. Omori [Milne, 1886; Lawson, 1910]. Milne experimented with base isolation and Omori with brick structures incorporating parabolic curves. After the Yokomana earthquake of 1880 Milne formed the Seismological Society of Japan and another temblor, the earthquake of 1891 spawned the Earthquake Investigation Committee. This committee published guidelines for seismically resistant wooden buildings [BEIC, 1900]. Some authors have proposed that even before the 18th century and 19th century wooden buildings had been built to incorporate seismically resistant features [Tanabashi, 19601. For example, Japanese pagodas are often built around a tall interior mast which hangs in the center supported by chains. As the pagoda oscillates in an earthquake the mast does not, acting like a damper to counteract drift. Sometimes the masts are fixed in the ground, helping to control excessive drift through their natural elasticity. Authors have also argued that the bracketing system imported from China helps distribute earthquake forces [Shiping, 1991]. But it is still unclear whether vernacular Japanese stores and residences were ever consciously designed to resist earthquakes.
Fire was a more urgent problem than earthquakes and fire-resistance added to the weight of Japanese roofs. As in the United States where brick buildings built to resist fires have become an enormous seismic problem, so in Japan where the mitigation of one danger has accentuated another. The Great Temnei fire of 1864 destroyed nearly eighty percent of Kyoto, a city of wooden buildings. The wooden dwellings and stores in present-day Kyoto, the most historic of all Japanese cities, are no more than 120 years old. The great Ginza fire of Tokyo in 1872 illustrated the danger of wood and spawned the brick Ginza district. The cities of Osaka and Kobe developed their own fire regulations which stipulated terra-cotta roofing in 1909 and 1912 respectively [Ohashi, 1992]. Fire was still a major problem in Tokyo after the earthquake of 1923 which burned through the traditional structures of that city. The almost universal use of terra-cotta roof tiles bedded in sand in the Kyoto-Osaka area seems a sound fire mitigation method but the flammable screens, mats, and small wooden members in building interiors could provide highly combustible fuel.
Only comparatively recently have regulations and codes been enacted to protect wooden buildings in earthquakes. The use of braces in new wood houses was first required in the Building Standards Act of 1950. Code provisions for bearing walls and minimurn ratios of wall areas to plan were introduced in 1960 and increased in 1981. Most likely the 1995 earthquake will spawn stricter more comprehensive codes.
This survey of earthquake damage to traditional Japanese stores and residences raises questions that call for further research. More historical research is necessary to be done to clarify the Japanese attitude toward earthquakes and to study how this attitude affected the design and construction of traditional structures. More testing of Japanese traditional houses and stores is necessary to understand their weaknesses and to suggest retrofit strategies. Even after the imposition of new code requirements in 1981, present-day construction could still be improved by the addition of more tiedowns straps and stirrups, better floor and roof diaphragms, and better lateral bracing systems. I saw many failures of recently constructed buildings. A careful retrofit strategy using artfully designed metal connections would improve the performance of the frame by securing joinery. Perhaps a lighter roofing system which had the beauty and fire-resistance of terra-cotta tiles could be invented to reduce the weight of roofs. It is clear that in order to save lives in the future traditional Japanese construction techniques must adapt to modern anti-seismic technology. The challenge is to propose solutions that will not prejudice the aesthetics of traditional buildings and that will not increase their already high cost. Hopefully a retrofit strategy can be devised before, these traditional buildings, so much a part of the Japanese architectural heritage, are lost.
Selected images of traditional wooden houses. Click on thumbnail for full-size photo.
[Tingey, 1981; Durston, 1987; Shimamura and Suzura, 1993]
William R. Tingey, “The Principal Elements of Machiya Design,” Process: Architecture (Tokyo, 1981), 25, pp. 83-102.
Diane Durston, Kyoto; Seven Paths to the Heart of the City (Kyoto, 1987).
Noboru Shimamura and Yukio Suzura’s Machiya of Kyoto (Kyoto, 1993).
[Itoh, 1974; Treib, 1976]
Teiji Itoh, Kura, Design and Tradition of the Japanese Storehouse (Seattle, 1974), pp. 31-35.
Treib, Marc, “The Japanese Storehouse,” Journal of the Society of Architectural Historians, May 1976, Vol. 35, No. 2, pp. 124 -137.
Preliminary Reconnaissance Report of the 1995 Hyogoken-Nanbu Earthquake (English ed.), The Architectural Institute of Japan (April 1995), pp. 53-56.
Andre Corboz in Henri Stierlin and Tomoya Masuda, Architecture of the World:Japan, (Lausanne, nd), p. 5
Heinrich Engel, The Japanese House---A Tradition for Contemporary Architecture, 1964, pp. 356-357.
Teiji Itoh, p. 35
Nfichio Fujioka, Kyoto-Gosho (Tokyo, 1956), pp. 169-170.
[Milne, 1886; Lawson, 1910]
John Milne, “On Construction in Earthquake Countries,” Minutes of proceedings of the Institution of Civil Engineers (London, 1886) 83, pp. 278-291.
Lady Lawson, Highways and Homes of Japan (London, 1910), pp. 250-258.
“Condensed Statement on the Construction of Earthquake-proof Wooden Buildings,” Bulletin of the Earthquake Investigation Committee in Foreign Languages (Tokyo, 1900),4, pp. 1-12.
Ryo Tanabashi, “Earthquake resistance of traditional Japanese wooden structures,” Proceeding of the Second World Conference on Earthquake Engineering at Tokyo and Kyoto, July, 1960 (Tokyo, 1960), pp. 151-163.
H. Shiping, “The earthquake-resistant properties of Chinese traditional architecture,” Earthquake Spectra, (1991), 7, pp. 355-389.
Yuji Ohashi, The History of Structural Codes of Buildings in Japan (Tokyo, 1992), pp. 51-52.
Updated December 9, 1997.
Send comments or questions to EERC Library