How Do Experts Work Out What a Lost Historic Building Looked Like Before Rebuilding It?

Rebuilders Layer Five Kinds of Clues: Foundations, Drawings, Photos, Models and Look-Alikes

Imagine you had to rebuild a house that burned down before you were born. You would probably start with the spot where it stood: where the walls were, how big the house was. Then you would ask relatives for old photos and floor plans. If one side of the house had never been photographed, you might look at the neighbor’s house, built by the same builder in the same year. Experts rebuilding a castle or palace follow nearly the same path.

The clues usually come in five kinds:

• What digging finds. Stone bases, post holes, roof tiles. These show where the building stood and how big it was.
• Drawings and surveys. Measured plans made while the building still stood.
• Old photographs. They show the outside, and sometimes the inside.
• Models and written records. A small wooden model made by the original carpenters, or a repair record.
• Look-alikes. Surviving buildings from the same period, used to fill in whatever is missing.

No building has all five in equal measure. The four examples below sit at very different points on that scale, which is why they are a good place to look.

The Suzaku Gate in Nara Borrowed Its Missing Upper Floor From Surviving Temple Buildings

The Suzaku Gate was the main south gate of the Heijo Palace, the emperor’s palace in Nara. The research institute that studies the site, the Nara National Research Institute for Cultural Properties, says the direct evidence was limited to three things: the foundation and its packed earth base, the positions of the posts (which give the floor plan), and the roof tiles that were dug up [1].

What about the rest? The institute says no material directly shows the gate’s structure. Instead, the team made choices based on other buildings. The gate at Heian, Japan’s next capital, had two stories, and the team took that into account when it gave the Nara gate two stories as well. For the basic frame the team followed the middle gate of Horyuji, a temple that holds the oldest wooden buildings still standing in Japan. For style they looked at a pagoda of Yakushiji temple, which is closer in date. Because the Suzaku Gate was so large, they checked the sizes and proportions of its timbers against a big gate at Todaiji temple. Small details came from a miniature pagoda, bells and roof ornaments dug up elsewhere [1].

The work took decades. The institute dates its research on the gate to the mid-1960s, when a 1/10-scale model was made after excavation, and a three-year study of design and structure began in the late 1980s [1]. The government’s regional office for the area lists the gate’s reconstruction as finished in 1998 [2].

One more detail shows how today’s rules shape such a project. In Japan a rebuilt gate counts legally as a modern building, so it had to be proved safe by modern engineering. The team hid diagonal braces inside the roof frame and stiffened the inside of the lower walls with wooden and metal framing [1].

So here the ground decided the plan, and surviving buildings decided the height. The upper floor you see today is a well-supported reconstruction, built on clues from other buildings rather than on direct evidence. That is the answer in miniature. It also raises a question: what if the old drawings and photos do survive?

Nagoya Castle’s Palace Had Survey Drawings and Photos, So Less Had to Be Guessed

Nagoya Castle in central Japan had a residence for lords, the Honmaru Palace. In 1930, together with the castle’s main tower, it was designated Japan’s first “National Treasure” of the castle-architecture category under the old system. It was lost to wartime damage in 1945. Nagoya’s city government says that, luckily, a large body of records had been kept safe: 309 measured drawings from the 1930s, about 700 old photographs, and 1,049 painted wall and sliding-door panels that survived the fire [3].

That is a very different starting point from the Suzaku Gate. Here, builders could work from drawings and photos of the building itself, rather than from similar buildings that survived. Those counts come from the city that runs the project, so they are the owner’s own description. It is still fair to say that the more direct records exist, the less a team has to rely on look-alikes.

Ozu Castle’s Main Tower Was Rebuilt From a Wooden Model and Photographs

Ozu Castle, on the island of Shikoku, lost its main tower (the “keep”) in the 1800s, when it was taken down. The castle’s official site says the town set out in 1994 to rebuild the keep in wood, and finished in 2004. The clues were a small keep model that is thought to date from the Edo period, and photographs taken in the Meiji era [4].

A model that shows the whole shape, plus photos that show the outside from several sides, can be checked against each other. The site describes it as one of the few keeps in Japan that can be rebuilt with accurate knowledge of its appearance [4]. That claim is the town’s own, so treat it as the owner’s view. The site also says the builders recreated old techniques, such as hand-forged nails [4].

At Shuri Castle, Experts Disagree on Which Old Record Should Decide Which Way the Dragon Pillars Face

Shuri Castle in Okinawa was the palace of the Ryukyu Kingdom, which ruled the islands before they became part of Japan. Its main hall was rebuilt in 1992 and burned down on October 31, 2019. A museum director said the 1992 rebuilding had to use pre-war materials gathered piece by piece, including records left by Yoshitaro Kamakura, because many records were lost in the Battle of Okinawa in 1945 [5].

The second rebuilding ran into a hard question: which way should the two big stone dragon pillars at the front stairs face? According to Ryukyu Shimpo, a 1768 repair record and an 1846 picture-book of work orders show them facing each other, while a photograph taken in 1877 by the French navy shows them facing the front [6].

In December 2021 the government’s technical committee decided, provisionally, on facing each other. Its chair said the committee valued the written records that it had already used for the building’s outside and inside, and judged that the direction had changed in the roughly 30 years between 1846 and 1877. But no clear proof of such a change has been found [6].

Critics push back. One scholar called the conclusion unscientific, since the one fact that is certain is that the pillars faced front in 1877. Researchers’ groups later delivered a petition with 6,047 signatures, arguing that the drawings show the front legs in the wrong places and that chisel marks on surviving fragments point to a front-facing pillar [6] [7]. Another expert argued that for this kind of work, an actual object should count for more than a drawing or a document [6].

At a briefing for residents, a committee member said that new finds could change the analysis, and the chair said that no answer can be 100 percent certain but that the committee would keep working from the materials, as Ryukyu Shimpo reported [8]. According to the newspaper Ryukyu Shimpo, the pillars were set in place facing each other in 2026, and the argument has not ended [9]. Because this is a news report, check an official announcement for the latest.

What this shows is that more evidence does not always make an answer. When the sources disagree, someone has to choose which kind to trust, and that choice is part of the rebuilding.

Other Countries Face the Same Choice, and International and Japanese Rules Tell Builders to Stop at Guesswork

None of this is only Japanese. Warsaw’s Old Town, in Poland, was reduced to rubble in 1944. It was rebuilt in the following years, largely to the way the city looked in the 1700s, with painted city views from that century among the records historians point to [10].

Experts have argued for a long time about how far rebuilding should go. A famous international text, the Venice Charter of 1964, says that restoration is based on “respect for original material and authentic documents,” and that it “must stop at the point where conjecture begins” (conjecture means guessing) [11]. Where exactly the line between a careful reading of evidence and guessing falls is what the arguments in this note are about.

Japan has its own rule. In April 2020 a Japanese government council for cultural properties set a standard for rebuilding historic buildings on protected sites. It followed an earlier report from 2017 and a 2019 working-group paper about rebuilding castle keeps. The standard narrows the range of rebuilding that counts as proper and sets out steps and points to watch [12]. I could read only the announcement, not the standard itself, so I do not describe its clauses.

One small thing to know if you read about Japan: the word for this work has two written forms, both pronounced “fukugen.” The Nara gate’s official pages use one, and Nagoya’s uses the other [2] [3]. In English, “restoration” and “reconstruction” are not used the same way everywhere, either.

You Can Read the Sign at a Rebuilt Building as a List of Clues

Next time you visit a rebuilt castle, palace or old town, try reading the information sign this way. Which of the five clues did the builders have: foundations, drawings, photos, models, or look-alikes? If the sign says nothing, the site’s office may be able to tell you. The answer says how much of what you are looking at is known and how much is an informed guess.

To follow a debate that is still going, look for updates on the dragon pillars of Shuri Castle from the committee or a trusted news source. You can watch experts weigh a drawing against a photograph, and you can decide which one you would trust.

Sources

Update history