Proportions of Wood Members in Japanese Traditional Architecture—A Comparison of the Kiwari-Sho and Measurements of Building Remains

Abstract

Since written reference works on ancient design techniques are still extant, it is possible to gain an understanding of the development of these designs. In particular, we are able to investigate dimensional standards used for wood members using the Kiwari (木割: a method of determining the dimensions of all wood members using a proportion determined based on the diameter of the columns) in Kiwari-sho (木割書: a manual of dimensional standards for wood members). The objective of this study was to confirm whether the design techniques that have been collected under the term “dimensional standards for building (木割術: Kiwari-jutsu)” were actually utilized in the construction of buildings and whether they are a sustainable technology. This study will provide an understanding of the correlation between the techniques used and the dimensions of the remains of structures putatively built using such techniques. Furthermore, based on this analysis, this study will be useful in both quantifying the influence that dimensional standards techniques had on buildings at the time of their construction and as reference material that can be used when restoring the remains of cultural properties.

1. Introduction

During the ancient age (before 784) of traditional Japanese architecture, a large number of wooden buildings were constructed that had similarities to ancient Chinese and Korean buildings due to the interaction between Japan and the two countries at the time. One such type of traditional architecture, temple-and-shrine architecture, began as a result of the spread of this kind of architecture in the ancient period. It later took on different characteristics that were developed over the passage of time and it still exists in the present day. Since Japan is one of the countries in East Asia in which written reference materials related to design technology are still extant, it is possible to gain an understanding of the development of this technology. In particular, it is possible to confirm aspects of technologies related to dimensional standards using Kiwari-sho (木割書: manuals of dimensional standards used for wood members). In China, one of the representative works on design standards is the 『営造法式: Yingzao Fashi』 (manual on building and construction methods).

In Japan, although research into Kiwari (木割: the dimensional standards for building) has led to the identification of design techniques related to methods used to determine intercolumn spacing, no studies have so far attempted to compare the Kiwari-jyutu (木割術: dimensional standards techniques) with actual measurements of dimensions. Thus, a study of design techniques that have been compiled since the Middle Ages to confirm whether these techniques were actually applied when constructing buildings in the past was utilized in this study. As this analysis will provide an understanding of the correlation between the techniques and the dimensions of the remains, it will enable the quantification of the degree of influence that dimensional standards techniques had at the time. While Kiwari-jyutu were techniques that originated in the Middle Ages of Japanese history, in the present study, an understanding of their origins is not important. Rather, since the objective of this study was to perform an analysis of the proportional relationships of wood members, important cultural properties were identified as the subject of this analytical study of actual dimensions.

The Kiwari-jyutu (dimensional standards techniques) included in the kiwari-sho have been continually studied by a variety of researchers, including Yōtarō Itō, Takeshi Nakagawa, Akira Naitō, and Katsuhiro Kawata. This study was concerned with the kiwari-sho and dimensional standards techniques in circumstances in which no studies comparing actual measurements of buildings with the dimensional standards techniques included in the kiwari-sho have thus far been conducted. Therefore, in this study, the extent to which dimensional standards techniques were involved in the actual construction of buildings was identified through a comparative study of the proportional relations of wood members based on actual measurements of buildings. It is believed that such a comparative study can prove whether this technology is sustainable.

This study targeted national treasures and important cultural properties whose basic design measures 3Ken (間: rooms) [1] each along the front elevation and the side elevation. The analytical resource consisted of the dimensions obtained from the published repair work reports on cultural properties. This was followed by a comparative analysis of those dimensions and the proportions listed in the kiwari-sho. The specific kiwari-sho used consisted of those that also indicate buildings measuring 3Ken on both the front and side elevations and whose documents can be dated.

Although Japanese kiwari-sho are texts that were written from the Middle Ages to the early modern period (1603–1867), the techniques listed in these works resulted from numerous attempts at construction performed prior to the Middle Ages. As a result, they are believed to also contain techniques that were developed prior to the Middle Ages. The items included in these kiwari-sho reference temple and shrine buildings measuring 3Ken per side. Considering that I conducted a comparison of the dimensions of the remains of such buildings, I analyzed the wood members that were used to construct the buildings by dividing them into four categories: floor plan (平面), foundation and lower supporting section (軸部 jiku-bu), the junction between the columns and the roof (組物部 kumimono-bu), and the roof (屋根部 Yane-bu).

1.1. Japanese Architectural Technology

The following is the theory behind the architectural technology that has been used in Japan up to the present day.

(1) Kansū-sei (完数制: traditional system of measuring proportions based on a standard figure, usually the shaku, which is approximately 30.3 cm).

One old method of design technology is believed to have been the “Kansū-sei for column distance” [2], in which the dimensions (measurements) for the distance between columns were calculated as the integer multiple of a unit of measure (a module);

(2) Shiwari-hō (枝割法: a system used to standardize the cutting of lumber).

Since the Middle Ages, relationships in Japan between the dimensions of columns, the distance between columns, the junctions between columns and the roof, and the Taruki (垂木: rafters) arose, which eventually became the method referred to as Shiwari-hō [3]. This method formed the basis for floorplan designs. The basic units used in the technique were the width of two Taruki and the distance between Taruki;

(3) Nakawaki-sa (中脇差).

Nakawaki-sa refers to a method of calculating the proportions of the floor plan of a building using an equidistant grid. It is the method of determining the distance between columns that was advocated by Kuninobu Ishii [4]. The basic unit of length (u) is determined, then the diameter of the columns (2u) is determined using the integer multiple, and then the grid (the distance n × u) is determined. The distance between the columns is then determined by placing three column measurements (inside measurement, center, and outside measurement) on this grid;

(4) Proportional method (比例法).

In Japan, research focused on the methods used to determine the distance between columns according to their proportions has been occurring for a very long time. The major methods include the “Theory of Determining Proportional Column Distance” [5], which uses √2:1 (≒1.414:1); the 3:2 ratio [6] based on the influence of a Zen Buddhist style advocated by Nobuo Itō; the 4:3 ratio [7] calculated using the dimensions of a structure known as Konjiki-dō at Chūsonji (Buddhist temple); and the 3:2 ratio [8] of Takuichi Takeshima;

(5) Kiwari (木割).

A summary of all the definitions of Kiwari would state that it is a method of determining the dimensions of all wood members using a proportion determined based on the diameter of the columns. Although there are differences based on the scale and style of the building, the proportions prescribed in each Kiwari-sho determine the diameter of the columns and the column diameters in turn determines the proportions of wood members such as the building’s footprint, the beams running between the columns, the Jihuku (地覆), Nagesi (長押), nuki (貫: joint connections), Keta (桁: purlins), Hari (梁: joist beams), and Taruki (垂木: rafters). Kiwari is an empirical method developed by artisans that has existed since prior to the Edo period (1603–1867) and it was recorded in a large number of Kiwari-sho.

The texts that contain the provisions for all wood members determined by the various styles and scales of buildings are known as Kiwari-sho, and the contents of each Kiwari-sho differ according to the carpentry lineage that created it. In simple terms, Kiwari-sho, which contain information related to Kiwari techniques, are a type of technical manual that were created from the Middle Ages and onward for the purpose of handing down to later generations the outstanding traditions of the carpenters of the age. Approximately 500 types of Kiwari-sho are still extant. Many of them were added to little by little by the successors of famous carpentry lineages. The contents of the 500 texts share some of the same guidelines. However, since they are old texts, some have not been discovered in their complete form, which means that the missing portions may be discovered at a later date.

In addition, Kikujutsu-sho (規矩術書) are an example of a kind of old text that is similar to Kiwari-sho. There are many similarities between kiwari and kikujutsu, but kikujutsu contains more detailed information than kiwari on numerical values for smaller wood members as well as joinery methods. Thus, strictly speaking, kikujutsu is a different type of technique. Put another way, kiwari-sho are texts used for the overall planning of a building, while kikujutsu-sho are texts that deal with joining and other parts of construction.

As mentioned above, the kiwari-sho that were the subject of this study were texts from the Edo period, which were studied to determine their sustainability over the ages. The validity of this study had to be confirmed in order to ensure its continuity.

1.2. Methods

The subjects of this study were National Treasures and Important Cultural Properties. A total of 16 National Treasures and 155 Important Cultural Properties were selected. Of these, 58 buildings have had survey reports published that contain all their measurements and dimensions [9] (Figure A2). The subjects of the comparative analyses were 46 items contained in 17 Kiwari-sho [10] (Figure A3). Kiwari-sho was written in “Kuzushiji (くずし字),” a language that is not used today, and therefore needs to be translated into modern languages (Figure 1). After translation, the translated contents were analyzed and between the wood members were arranged to become the Kiwari-jyutu proportional equation. Comparison of this proportionality formula with the actual dimensions of the remains was then conducted.

The multiple of integers utilized in this paper are equal multiples. In particular, a margin error of ±0.1× was applied to multiples of 1.5, 2×, and others. Multiples of 1/3 to 1/9 were applied within the integer multiples without the use of any margin of error [12].

Based on the above analysis, the comparisons in this study were performed using the ChuoMa, WakiMa (distance between columns), and Mokosi (pent roofs) shown in the floor plans. Jiku-bu (Measurements taken from the supporting structures); Kumimono-bu (the junctions between the columns and roof); and Yane-bu (the roof) wood members included the diameters of pillars; RaikoHasira (round pillar diameter); Nuki (joint connection) height and width; KasiraNuki (the joint section) height and width; Daiwa (the architrave) height and width; Hijiki (bracket arm) length and width; Daito height, length, and width; Makito height, length, and width; Gagyo (round purlins) height and width; and Taruki (rafter) height and width [13] (Figure A4 and Figure A5).

2. Analysis According to Each Part

2.1. Floor Plan (平面)

Figure 2 shows the multiple relationships between the dimensions of the ChuoMa, the WakiMa (中央間, 脇間: distance between columns), and the Mokosi (裳階: pent roofs) based on the diameter of pillars and the Issi (一枝) [14]. The light-colored area in the center of the shaded portion shows the integer multiple, while the dark-colored area shows the 1/3–1/9 multiples, which are the equal divisions.

The analysis of the ChuoMa based on Figure 2 showed that the integer multiples calculated based on the column diameters were found in 25 locations on the front elevation, 24 locations on the side elevation, and in 17 locations on both elevations. Equal multiples were found in three, two, and two locations (in the order of center, front, and side). In the WakiMa, there were 24, 23, and 21 locations, respectively, and equal multiples in 2, 2, and 2 locations, respectively. In the Mokosi portion, there were six and six locations, respectively, and all the locations were both integer multiples.

Shiwari was used as a standard measure to analyze the proportions of the floor plan. The integer multiples in the ChuoMa were present in 32 locations on the front elevation, 31 locations on the side elevation, and 29 locations on both elevations. In the WakiMa, there were 27, 26, and 25 locations (in the order of front, side, and both) and the equal multiples were in 1, 1, and 1 location, respectively. In the case of the Mokosi, there were five locations on the front elevation and five locations on the side elevation.

The results of this analysis showed that the floor plan proportions were determined using Shiwari rather than the column diameters. It is likely that Shiwari multiples were used as the standard module rather than Shiwari applied to the dimensions.

Four buildings had a standard Shaku (尺: 303 mm) as their column diameters, and a multiple of a shaku as the distance between the columns. Repairs are believed to have been made in the early modern period based on the measurement of a Shaku [15].

No relationship was found between the sizes of the floor plans and the integer multiples of the column diameters in these buildings. There was no correlation with the integer multiple in spite of conducting analyses based on regional differences and the time periods in which the buildings were constructed.

Comparison of the proportions of the ChuoMa and WakiMa making up the floor plan and the proportions of the column diameters with the Kiwari proportions (Figure 3) showed that many of the building plans for the ChuoMa used the same dimensions for the front and side elevations and that there were 16 and 9 locations, respectively, in which the proportions were the same as the 0.1D [16] and 0.11D of the Kiwari-jutsu. The 0.12D was found in only four locations and 0.15D was found in none, indicating that these were few in number. The 0.08D and 0.09D, which are not part of Kiwari-jutsu, were found in nine and five locations, respectively, indicating that, in comparison to the overall distance between columns, the column diameters were smaller. On the side elevation, 0.1D, 0.11D, 0.12D, and 0.15D were found in 18, 11, 4, and 1 location, respectively, indicating concordance with kiwarijutsu. In the case of the peripheral room, in the front elevation, the 0.12D, 0.15D, and 0.16D proportions were used in 9, 10, and 4 locations, respectively, while in the side elevation, they were found in 10, 12, and 4 locations, respectively.

Analysis using Shiwari on the same ChuoMa, WakiMa, and Mokosi portions showed that, first, from the perspective of proportions, there were 13 locations (buildings) using the Mabaradaruki (疎垂木) equidistant placement system. It is thought that this was done in consideration of the scale of the buildings (yellow portion in the Figure 3). Many of the others did not match the Shiwari of Kiwari-jutsu.

An investigation of the floor plan designs showed that many utilized multiples of Issi rather than the column diameters, and that the Kiwari-jutsu corresponded to the proportions derived from multiples based on the column diameters.

2.2. Jiku-Bu (軸部: Supporting Structure)

The wood members that formed the Jiku-bu were the RaikoHasira (来迎柱: round pillars), Nuki (貫: joint connections), KasiraNuki (頭貫: joint sections), and Daiwa (台輪: architraves). Analyses of each of these based on column diameters are shown in Figure 4. In the case of the RaikoHasira, 19 of the 44 buildings had integer multiples derived from the column diameters. Nuki heights were derived from the column diameters in 20 out of 57 buildings, and the Nuki widths were derived from the column diameters in 11 out of 56 buildings. In addition, the heights of the KasiraNuki were derived from the column diameters in 22 out of 54 buildings, as were the widths in 28 out of 52 buildings. In the case of the Daiwa, few of the buildings had these structures, but there were 4 locations out of 23 in which the height was derived from the column diameters and 11 out of 24 in which the width was also derived. Consideration of the overall percentages that these figures represent indicated that the building material dimension with the most commonly used integer multiples was the KasiraNuki width (54%), which is correlated with the column diameters because it is a structure that pierces the column. However, the KasiraNuki structure, which was linked to the columns in the same manner as the Nuki, comprised the lowest percentage (20%). The likely reason for this is that the KasiraNuki was smaller than the Nuki (joint connection).

Analysis of the wood members with integer multiples based on Shiwari indicated that for RaikoHasira, there were 17 out of 40 buildings; for Nuki height, there were 13 out of 49 buildings; and for Nuki width, there were 13 out of 48 buildings. KasiraNuki height was registered to be in 19 out of 47 buildings and KasiraNuki width was used in 14 out of 44 buildings. Daiwa height was registered to be in 7 out of 20 buildings and Daiwa width was used in 12 out of 20 buildings. Overall, few wood members used integer multiples, and there was little correlation with Shiwari. Observation of the percentages based on these results showed that fewer than 43% of all wood members used this method, although Daiwa width used it in 60% of buildings. The Daiwa is a structure used in Zen-style (one of the Japanese architectural styles: a Chinese-style architecture) buildings. Based on its relation to the Kumimono (junction) between the columns and the roof, it is believed that it utilizes Shiwari and integer multiples, but that determination was made in conjunction with results obtained from analyses of the junction between the columns and roof. The lowest values were Nuki height and width, both at 27%, which did not match the results obtained from Shiwari.

An analysis of the proportional relationship of the wood members was conducted by analyzing the proportions of the wood members based on Kiwari-sho and those of the wood members obtained from the dimensions of the wood members in actual buildings (Figure 5). An analysis of the Kiwari-jutsu for the RikoHasira showed there were nine and five locations, respectively, that used 1.2D and 1.3D. There were 14 and 8 examples using 1.1D and 1.0D, respectively, which are not part of Kiwari-justu, showing that these were used in larger numbers of buildings. An investigation of the proportions of the RikoHasira showed that the dimensions for the distances between columns were not consistent even within the same proportional method and that the dimensions within the same proportions for column diameters were likewise not consistent.

Based on the above results, when I observed the tendencies shown by the horizontal structures (Nuki) other than the RaikoHasira of the supporting structure, I found that they were 0.1 times larger.

The results of proportion analysis showed that of all building material proportions and the percentages that are in accordance with Kiwari-jutsu, overall, only a small percentage of actual buildings were in accordance with Kiwari-justu. Over 50% of the buildings had Daiwa, and the heights of the Nuki, the widths of the KasiraNuki, and the heights and widths of the Daiwa were 0.1 times larger than Kiwari-jutsu.

2.3. Kumimono-Bu (組物部: Junctions between the Columns and Roof)

Investigation of the Kumomono-bu included the Hijiki (肘木: bracket arm), Daito (大斗), and Makito (巻斗). Analysis of these wood members using the column diameter as the standard is shown in Figure 6. First, when one observes the results that correspond to the integer multiples derived from the column diameters, one can see that the height of the Hijiki matched the building material dimensions in 28 out of 53 buildings and the widths of these structures matched the building material dimensions in 13 buildings. An analysis of the Daito showed that, out of 44 buildings, the height matched in 25 cases, the width in 24 cases, and the length in 21 cases. Analysis of the Makito showed that, out of 42 buildings, the height matched in 20 cases, the width in 21 cases, and the length in 27 cases. The match rate for the Hijiki was low, at 24%, while that of the Makito length was high, at 64%. An analysis of the integer multiples for the Kumimono-bu based on Shiwari showed that there were matches for the Hijiki heights in 16 cases, for the Hijiki widths in 12 cases, for the Daito heights in 18 cases, for the Daito widths in 21 cases, and for the Daito lengths in 19 cases. In addition, there were matches for the Makito heights and widths in 13 cases each and the lengths in 17 cases. Examination of the match rates showed that only the Daito width had a rate that was over 50%. Overall, the rates were under 50%, with a supporting structure consisting of wood members that support the Taruki (rafters) at a low percentage.

Analysis comparing Hijiki heights with Kiwari-jutsu (Figure 7) showed that there were 9, 21, 12, and 15 examples, respectively, that matched the Kiwari-jutsu measurements of 0.3D, 0.4D, 1.2 width, and 1.3 width, while there were 13 and 11 examples, respectively, that measured 0.5D and 1.1 width, which are not part of kiwarijutsu. In the case of width, 20 and 28 examples of Daito widths matched the kiwarijutsu measurements of 0.3D and 1/3 width, respectively. Non-Kiwari-jutsu measurements found 3 examples of 0.2D, 16 examples of 0.4D, and 13 examples of 0.4 Daito width. Based on these results, I found that the heights of Hijiki were 0.4 and 0.5 times the column diameter, and the widths were 0.3 and 0.4 times the column diameter.

Analysis of Daito height showed that there were 18 and 13 examples, respectively, of 0.5D and 1.5 Hijiki height, which matches Kiwari-justu, while there were 3, 11, 7, and 2 examples, respectively, of the non-Kiwari-jutsu 0.4D, 0.6D, 1.4, and 1.6 Hijiki height. The 1.5 Hijiki height for Kiwari-jutsu was 0.3D when calculated using the proportional method; 0.45D and 0.6D when using the 0.4D method; and 0.36D and 0.39D, respectively, when based on 1.2 Hijiki width and 1.3 Hijiki width. Therefore, the Daito height was determined to be 0.4D, 0.5D, and 0.6D. In the case of width, the proportional method of Kiwari-jutsu resulted in 0.8D and D, of which there were 6 and 12 examples, respectively, while there were 7 and 5 examples, respectively, of the non-Kiwari-jutsu measurements of 0.9D and 1.1D. Length determinations were 0.8D and D, as was the case with width. The number of matches was the same as with width.

Investigation of Makito height showed that the determinations made using Kiwari-jutsu were Hijiki height and 0.9 Hijiki height and that 19 and 7 examples, respectively, matched these measurements. Investigation of the 1.1D proportion found that there were 8 examples. In the case of width, there was a wide variety of dimensions, with 4 examples matching 5/3 Hijiki width and 0 and 7 examples, respectively, when calculated using 0.3D and 0.4D. In the analysis of the length measurements in accordance with Kiwari-jutsu, 13, 6, 7, 11, 7, and 6 examples, respectively, matched 4/6 Daito width, 2 Taruki width + between Taruki, and Taruki, 0.48D, 0.52D, 1.2 Hijiki height, and 1.3 Hijiki height. There were few (six) examples of 2 Taruki width +between Taruki and Taruki, which were related to Shiwari. There were seven and six examples, respectively, of 1.2 times and 1.3 times, which were measurements based on Hijiki height. Measurements not determined this way included two examples of 1.1 times and eight examples of 1.4 times.

Based on the above results, the percentages of the whole that were made up by Kiwari-jutsu show that the measurement with the highest match rate was the Hijiki, with a match rate of 70% for height and 57% for width. The heights of Daito and Makito had percentages of over 60%, but the widths and lengths had low percentages. Looking at the overall results for the Kumimono-bu, the vertically oriented wood members utilized Kiwari-jutsu while the horizontally oriented wood members did not.

2.4. Yane-Bu (屋根部: Roof)

Analysis based on the wood members that made up the roof—the Gagyo (丸桁: round purlins) and Taruki (垂木: rafters) (Figure 8)—showed that the Gagyo were included in the reference materials but were mentioned only rarely. Thus, analysis of Gagyo using materials for 25 buildings in terms of height was conducted. There were 11 examples in which the integer multiples were based on column diameter and 14 examples in the reference materials for the 24 buildings related to Gagyo width. Based on the analysis of the reference materials for 57 buildings, there were 10 examples of Taruki height and 5 examples of Taruki width. Few of the integer multiples were based on column diameter. Thus, it was determined that there was little correlation between column diameter and Taruki size. Analysis using Shiwari showed that there were 9 examples of Gagyo height in the reference materials for 19 buildings and 5 examples of Gagyo width in 18 buildings. An examination of the Taruki showed that integer multiples were used in 20 examples for height and 14 examples for width in the 50 buildings investigated.

Figure 9 shows the results of comparisons with Kiwari-jutsu. Examination of the height of the Gagyo indicated that there were 5, 6, 4, and 4 examples, respectively, that matched the following Kiwari-jutsu measurements in 25 of the target buildings: 0.5D, 0.6D, 0.7D, and 0.8D. Investigation of width showed that there were 2, 10, and 2 examples, respectively, of 0.25D, 0.4D, and Makito width measurements as well as 2 examples of 0.9 Makito width. Taruki height presented 11 and 9 examples, respectively, that matched 1.2 (1.25) and 1.3 measurements. Investigation of other proportions (1.0, 1.1, and 1.4 width) found that there were six, seven, and five examples, respectively, of each, indicating inconsistency. Width was 0.2D with 33 examples matching, while there were 17 examples that matched the non-Kiwari-jutsu 0.3D.

These results indicated that Gagyo height matched Kiwari-jutsu measurements in over 70% of cases and that there were many widths in the 0.4D proportion. Investigation of Taruki showed that there was little correlation between column diameter dimensions and integer multiples. As for height, there was great inconsistency in the measurements, and width was determined using both the Kiwari-jutsu measurement of 0.2D and the non-Kiwari-jutsu measurement of 0.3D.

3. Analysis Based on the Matching Rates between Kiwari-Jutsu and Actual Measurements

Analysis of the usage of Kiwari-jutsu proportions in the wood members of the 58 buildings targeted in this study showed that 21 buildings had at least 16 matches in the 31 analytical items used in the comparisons. It was further identified that those 21 buildings were constructed between the 15th century and the early 17th century. The analysis according to region showed that there were three buildings in Fukushima, one in Saitama, three in Chiba, one in Aichi, one in Ehime, one in Hiroshima, one in Aomori, one in Kanagawa, and one in Oita. This indicates that many of the buildings were located in the Tōhoku and Kantō regions of Japan and that there was a particularly high kiwarijutsu match rate in Chiba.

Based on the match rates for all wood members, it was found that those that had a match rate of at least 50% in the target buildings were ChuoMa (central room) on the front elevation and side elevation, Nuki (joint connection) width, KasiraNuki (joint section) height, Daiwa (architrave) height and width, Hijiki (bracket arm) height and width, Makito height, Gagyo (round purlin) height, and Taruki (rafter) width. Analysis of floor plans showed that there were many buildings with 3Ken structures in which the lengths of the peripheral rooms and the ChuoMa were the same. If the proportional method of Kiwari-jutsu for ChuoMa is applied to WakiMa, the matching rate becomes the highest among all materials. These results indicate that there is a high probability that the floor plan designs were made using Kiwari-jutsu. However, overall, in the case of buildings in which Kiwari-jutsu was used in at least two of the measurements for wood members, no correlation was found with the size of the building remains in which the proportional method was Kiwari-jutsu. In addition, there was no fixed format for the proportional formula and the size of each member in the applied Kiwari-jyutu.

4. Conclusions

The conclusions and the main findings of this study can be formulated as:

1. The analysis of floor plans showed that the ChuoMa, WakiMa, and Mokosi had an integer relationship to Shiwari rather than column diameter and did not match the Shiwari included in Kiwari-jutsu. Analysis comparing Kiwari-jutsu to the remains showed that there was a tendency for proportions to be smaller than the Kiwari-jutsu proportions. Some column diameters were based on a standard Shaku (303 mm), and the distance between columns was based on the integer multiple of that figure. However, it was also found that all other wood members had no proportional relationship to that multiple;

2. It was found that over 50% of the buildings had Daiwa and that the heights of the Nuki, the widths of the KasiraNuki, and the heights and widths of the Daiwa were 0.1 times larger than Kiwari-jutsu;

3. An analysis of the Kumimono-bu showed that the vertically oriented wood members utilized Kiwari-jutsu while the horizontally oriented ones did not;

4. The Gagyo of the roofs had a high match rate with Kiwari-jutsu, and the Taruki had a low match rate with column diameter and Kiwari-jutsu;

5. There were 21 buildings that had Kiwari-jutsu proportions. These buildings dated to between the 15th and early 17th centuries. The regional analysis results showed that many of the buildings were in the Tōhoku region of Japan and that buildings in Chiba had a high match rate with Kiwari-jutsu. Based on the match rates for all types of wood members, there were 11 examples in which there were matches for at least 50% of the 19 items. Of these, it is highly likely that the floor plans—which are an important part of the composition of the buildings—utilized Kiwari.

Analysis of the dimensions of architectural remains showed that Kiwari-sho, which are technical manuals from the Japanese Middle Ages, were used to design the buildings studied. In addition, based on the fact that I found no relationship between the regions and historical periods investigated, I believe that the dimensions of these architectural remains were changed during later restorations. Comprehensive assessment of the extant texts in Japan known as Kiwarisho showed that they record a method that is sustainable.