Hirosaki Castle: A History of Staying Upright

In 1897, the master carpenter Horie Sakichi moved Hirosaki Castle's keep westward, away from damaged stonework. Repairing the wall took much longer: that work was completed in 1915. For a time, keeping the building meant separating it from the place where it belonged.^[1]

The episode preceded the much-publicized move of 2015 by more than a century. It also exposes something easily concealed by the idea of an original castle. A building and its supporting ground have different histories. One may be serviceable when the other needs extensive work. The same is true of a wooden frame and its roof covering. Their association in a familiar silhouette does not make them a single, equally durable object.

Hirosaki's survival can be followed through these unequal lifetimes. Its walls, roofs, gates, and tower required different responses to failure. The most revealing history is often found where one part was altered so that another could remain.

Tsugaru Nobuhira completed Hirosaki Castle in 1611, carrying forward a project associated with his father, Tamenobu. This seat of the Tsugaru domain in northern Honshū comprised six enclosures separated by earthworks and moats. Its western moat followed a former course of the Iwaki River. The fortified landscape was much larger than the building now most readily recognized as the castle.^[1]

Its stonework accumulated over time. The central part of the eastern side of the honmaru, the main enclosure, initially had an earthen slope above low basal masonry. Work completed in 1699 extended the stone facing. The imposing edge beneath the later keep was itself a revision to the founding landscape.^[9]

The first keep, a five-story building in the southwest, burned in 1627. The surviving three-story tower was built in 1810 under Tsugaru Nobuchika, who obtained approval for a turret from the shogunate, the Tokugawa military government. It occupied the southeast corner instead. Construction date, location, and official description all distinguish it from the lost tower.^[2]

The change in corner matters. The later builders were not filling a vacant outline left by the fire. They placed a smaller building into a different relationship with the surrounding enclosures. The city's architectural catalogue describes its more elaborate gables and projecting features on the east and south, facing the ninomaru, or second enclosure; its inward-facing elevations are plainer.^[2] The tower addressed different audiences differently. Its public image depended partly on which side someone could see.

Gates and other turrets preserve more of the castle's working extent. The three surviving ninomaru turrets belong to an earlier building tradition than the keep and have their own repair histories. Five surviving gates mark passages between fortified spaces.^[3][4] These buildings supplied watching positions and controlled crossings across a much larger site, rather than forming decorative satellites around the 1810 tower.

A single construction date consequently gives a misleading starting point for maintenance. The keep, gate frames, roofs, and wall beneath them had already undergone different sequences of use and replacement before modern conservation began. Saving the castle would require deciding which parts of those sequences to carry forward.

At Hirosaki, snow clearing belonged to the work of government. The local architectural history assigns paths, riding grounds, and roof clearing to the domain's cleaning office, which could hire townspeople when additional labor was needed. Clearing the enclosed approaches outside gates fell to the foot soldiers stationed in their guardhouses.^[5]

The division followed the way the castle was used. A route could remain blocked after a roof had been cleared; a guarded gate could be standing yet difficult to reach. Maintaining access required work across the boundaries that the fortifications imposed. The labor of using a castle continued in weather that had no regard for those boundaries.

Material choices addressed another problem. The same architectural account links frost-damaged fired tiles and resulting leaks with the adoption of copper-covered wooden roofing. Copper was wrapped around wooden forms, preserving the tiled appearance through a different combination of materials.^[5]

This was not a choice between a completely wooden roof and an entirely metal object. The form beneath the covering and the surface exposed to weather performed different tasks. A roof could acquire a familiar tile profile without depending on fired ceramic for that profile. The innovation lay partly in separating shape from weathering surface.

Nor did all roofs follow one uniform specification. The city's turret catalogue records former split-wood shingles and the later use of copper worked to resemble that covering.^[3] Similar metal seen from the ground can conceal different roofing histories. The castle's apparent visual coherence is the result of adaptations made to several buildings, not proof that one original roofing campaign supplied them all.

Snow removal and reroofing also operated on different timescales. Clearing dealt with an accumulation already present. Changing materials attempted to reduce recurring damage, while leaving the need for care intact. Neither depended on the castle facing an enemy. A domain maintaining its headquarters had to keep entrances passable and water outside its buildings during ordinary years of administration.

One gate now provides unusually precise evidence of how that maintenance became part of the fabric. Its roof has preserved dates that were hidden from everyone using the passage below.

During repairs reported in March 2026, workers examining the Ninomaru East Gate found ink inscriptions dated 1815. They occurred on different roof components, including a copper element and wood beneath the ridge. The city's report connects them with an entry in the domain's official diary recording repairs to this gate in the same year. It also identifies a 1754 decision to adopt copper-covered wooden roofing progressively.^[6]

The written record and the hidden marks meet at a specific building. An administrative entry can establish that work was recorded without showing how much of it remains. A dated piece of roofing establishes a material connection, while needing the building record to explain the campaign to which it may belong. Together they make the early nineteenth-century repair more substantial than either an isolated date or an unattributed old component.

Later work did not necessarily remove that connection. A June 2026 conservation report states that mid-twentieth-century repairs had reused copper dating from 1815 and subsequent periods. It also records damage both to metal and to the wooden material beneath it.^[7] What looked like one roof contained earlier work retained through another repair campaign.

Whether one old component should go back into the roof was debated. In March 2026, officials proposed repairing and reusing the west oniita, the decorative ridge-end board dated 1815. Committee comments favored considering storage instead: renewed decay inside its copper covering could destroy the inscription. They requested careful recording even if reuse went ahead. The recorded conclusion called for preserving and making use of the three inscribed pieces, without equating preservation with their return to the roof.^[8]

The disagreement concerned two kinds of survival. Reuse would keep the component within a working building. Removing it for safekeeping would separate it from that function while protecting its value as a document. Its physical capacity to serve again was therefore only part of the question. The writing had given the material another responsibility: carrying evidence of an earlier repair.

An eighteenth-century decision and a nineteenth-century inscription had become relevant to a twenty-first-century building problem. The gate's custodians inherited both an architectural form to maintain and evidence that might be better protected outside it. Keeping the history and keeping every piece in use could lead toward different choices.

Stone presented a different kind of maintenance problem. A damaged roof component might be taken down while most of the building stayed in place. Work beneath the keep involved the ground supporting the whole structure. The rescue of 1897 separated those two tasks: moving the timber building reduced its exposure to the failing wall without repairing the wall itself.

By then, the castle was under a different system of care. The Meiji authorities had abolished the domains, and the grounds had passed through military control before opening as a public park in 1895. The major wall failure followed in 1896. When reconstruction finally proceeded in 1915, the contractor was Horie Hikosaburō, Sakichi's son. The works preceded the Taishō emperor's visit for military maneuvers.^[9]

The change of custodian did not make the physical problem new. A park had inherited the high enclosure edge and the building above it. Those features no longer served a ruling domain in the same way, but they still required support. The interval between rescue and repair shows that protecting a movable structure and restoring a large retaining wall were separate undertakings, with separate practical requirements.

Later investigations found the 1915 repair inside the wall: concrete near its basal stones and additional masonry behind the visible face. These were attempts to strengthen the structure, concealed within something that continued to look like an old castle wall.^[5]

This inner construction changes what an old surface can mean. Retained stones can stand in front of work much newer than themselves. A repair may seek continuity of appearance while changing the means by which that appearance is supported. The distinction is particularly consequential beneath a timber tower: the building's survival depends on what the wall does, not only on what its face looks like.

A century later, the keep was moved again so that another repair could reach beneath it. This time, dismantling also created an opportunity to investigate the wall as a historical structure, including the earlier attempts to keep it standing.

The 2015 move carried the standing keep about seventy-eight meters into the main enclosure. The historic timber structure was moved intact; the project did not substitute a new replica for it. In the subsequent wall dismantling, workers numbered individual stones and removed 2,185 of them while examining the masonry and its backing. Resetting the wall's stones was completed in December 2024.^[10]

Numbering changed how the dismantled material could be handled. A pile of stones preserves material but loses arrangement. Identifiers allow observations about a particular piece to remain connected with its place in the structure. The wall could be taken apart without reducing every stone to an interchangeable supply item.

This matters because a retaining wall is a relationship among pieces, ground, and the material behind its face. The visible stone provides only one part of the explanation. Dismantling exposes connections that cannot be examined in the same way from the moat, while also disrupting them. Recording becomes part of the intervention rather than something to do after the engineering work is finished.

The city's archaeological report describes investigations extending around and beneath the lowest stones. They identified earlier low masonry and traces associated with the nineteenth-century collapse.^[9] The wall was disclosing earlier forms of itself. Some evidence belonged to construction, some to failure, and some to the work that had followed failure.

The findings made the familiar boundary less chronologically simple. It contained material from a castle that had changed its enclosing surfaces, suffered collapse, and acquired reinforcement. Returning every part to a supposed single founding appearance would discard some of that history rather than recover it.

Moving the keep had therefore opened more than a construction site. It made a portion of the castle available for study that the keep's continued presence had helped conceal. That benefit came with a responsibility: decisions about the rebuilt support would affect the old ground now being examined.

The modern foundation scheme takes the separation of building and wall a step further. Four reinforced-concrete piles, each approximately thirty-five meters long and two meters in diameter, are designed to carry a platform beneath the keep. The city explains the purpose directly: the tower's weight will no longer bear directly on the stone wall.^[10][11]

The load is being given a different route into the ground. The retained tower and restored wall will recover their familiar spatial relationship, while the new support system carries a task that their visible association might still suggest belongs entirely to the masonry. The scheme preserves that association without requiring the wall to perform precisely its former structural role.

Preparing the pile locations also required archaeology. A municipal report published in July 2025 describes excavation within steel-lined shafts before deeper mechanical work. The previous year's investigation had identified part of a late Jōmon settlement, approximately 2,600–2,400 years old, as well as postholes predating the Edo-period tower platform.^[11]

A project undertaken to protect an 1810 building had encountered evidence much older than the castle. The discoveries were not an unexpected earlier version of Hirosaki's keep. They belonged to other uses of the same ground. The new foundations had to be understood as an intervention into that longer sequence, not merely an addition beneath a freestanding architectural object.

Construction itself continued to revise the working assumptions. The March 2026 committee minutes record boulders encountered during pile work and leakage of drilling fluid toward the moat. Work was suspended while changes to the method were considered.^[8] The planned geometry of a foundation could be drawn in advance; the behavior of the particular ground still had to be confronted during execution.

These records show the distance between a completed design and a completed repair. Archaeological investigation, temporary support, pile construction, and the movement of the tower each addressed a different problem. As of 16 September 2026, the return and foundation works remained in progress. The latest reviewed official report, describing conditions at the end of August, distinguished completed lower bearing work from upper components still to follow.^[12]

The new system is consequently part of an ongoing conservation history, not a demonstrated guarantee that the castle will need no further work. Its distinctive choice is already clear: keeping the old tower includes changing how it is supported.

In August 2026, the return operation supplied an unusually tangible form of public participation. During the weekend of 22–23 August, participants pulled four ropes as part of the controlled moving operation. The city's report records 2.22 meters of movement over those two days. Further sessions involved other participants and local schoolchildren.^[13]

A few meters are a modest distance across a park, but a substantial event when the object moving is its castle tower. The activity made separation between building and site visible to people taking part. For that interval, the landmark was something whose position could change through organized work, rather than an immovable point against which other movement was measured.

The participation did not complete the return project, and it did not make the public responsible for designing its foundations. It gave one part of a long technical undertaking an accessible scale. The stone numbering, concealed roof dates, and deep excavation required different explanations because they could not all be communicated through the same act of pulling.

Taken together, Hirosaki's records reveal several ways of keeping old material: repair a usable roof component, move a threatened building, record a wall before dismantling it, or provide a new route for its load. The appropriate action changes with the part being protected. Preserving an original does not necessarily mean preserving its position at every moment, or every feature of the system on which it once depended.

Hirosaki's historical substance lies in those specific continuities. A dated roof piece can be kept as evidence even when it no longer returns to its roof. An 1810 tower can remain because it is moved and given new support. The castle's familiar outline depends on decisions made beneath the copper, behind the stone, and in the ground below.