This is Volume I and the first volume of The Complete Raumschach Theoretical Series (IRF, 2026). It is offered as the most intimate work of the series — a book not about positions or plans or openings, but about the pieces themselves.
The five succeeding volumes of this series address Raumschach as a whole — the principles that govern it, the phases through which it passes, the plans and patterns that decide it. This first volume takes the opposite approach. It zooms in rather than out, narrows rather than broadens, and asks not “how does the game work?” but “how does this piece work?”
In flat chess, great players have always spoken of pieces with something approaching affection and individuality. Nimzowitsch loved the eccentricity of the knight — its ability to leap over other pieces, its awkward range at the edges of the board, its compensating power at the center. Tal loved the aggression of the rook on open files. Fischer had a famous preference for the bishop pair and wrote about it with geometric precision. These relationships between player and piece are not sentimental — they reflect the deepest kind of chess understanding: the knowledge of what a piece wants, where it is happy, and what it can accomplish that nothing else can.
Raumschach has six piece types, one of which — the Unicorn — is entirely without precedent in any other form of chess that has ever been played. Each piece deserves a formal monograph: a structured study of its movement geometry, its value, its opening role, its middlegame function, its endgame power, and the characteristic patterns it generates. That is what this volume provides.
Read each monograph slowly. Set up the positions on a board. Move the piece through its ranges and feel how the three-dimensional space opens or closes around it. The understanding gained from this kind of intimate acquaintance with individual pieces is different from — and complementary to — the strategic understanding of the succeeding volumes. It is the difference between knowing the rules of harmony and knowing the sound of a cello.
| Starting count | 10 per side |
| Move directions | 2 — forward or upward |
| Capture directions | 5 — across all planes |
The Raumschach pawn is the piece most transformed by the third dimension — not because its movement has become more powerful, but because it has become more complex. A flat-chess pawn has exactly one non-capture movement direction (forward one rank) and two capture directions (diagonally forward). The Raumschach pawn has two non-capture movement directions and five capture directions.
The two non-capture movements for a White pawn at position (L, f, r):
The five capture directions for a White pawn at (L, f, r):
The upward-forward capture is the most important and most frequently overlooked. A White pawn at Bc2=(2,3,2) can capture at Cc3=(3,3,3) — the absolute center of the board — via this direction. No other first move by any other piece reaches Cc3 in one move. The upward-forward pawn capture to the absolute center is one of the most powerful available opening moves in the game — except that it is a capture, and requires an enemy piece to be on Cc3 first.
Every Raumschach pawn has two ambitions: to advance in rank (toward the opponent’s home level) and to ascend in level (toward Level E). The art of pawn management is knowing which ambition to pursue at each moment — and the answer, as a general principle, is: ascend first, then advance. A pawn on Level C rank 2 is more valuable than a pawn on Level A rank 5, because the Level C pawn is in the strategic heart of the board.
A White pawn must reach Level E, rank 5 to promote. Starting from Level A or B, rank 2 or 3, this requires a minimum of seven moves — the longest promotion journey in any flat chess configuration. Passed pawns in Raumschach are therefore more valuable per square advanced than in flat chess.
From Ac2 = (1, 3, 2): Route A (ascend first): ♙︎Ac2–Bc2–Cc2–Dc2–Ec2–Ec3–Ec4–Ec5 ✓ [7 moves] Route B (advance then ascend): ♙︎Ac2–Ac3–Bc3–Cc3–Dc3–Ec3–Ec4–Ec5 ✓ [7 moves] Route C (mixed): ♙︎Ac2–Bc2–Bc3–Cc3–Cc4–Dc4–Dc5–Ec5 ✓ [7 moves] All three routes: 7 moves minimum from Level A, rank 2.
Raumcapa values advancement on a curve rather than a line, which is the right shape for the claim just made. Its passed-pawn bonus scales with the square of how far the pawn has advanced: a passer at advancement 4 of the maximum 8 earns +48 cp, while one at advancement 8 — a single step from promoting — earns +192 cp, four times as much for twice the advancement. A pawn three-quarters of the way home is worth far more than three-quarters of a promoted Queen’s value would suggest, and a quadratic curve is the simplest function that captures it: danger compounds as the remaining distance shrinks, it does not merely accumulate.
All flat chess pawn structures exist in Raumschach, plus three-dimensional analogues:
Rank-Doubled Pawns: two own pawns on the same level and file, one rank apart — the direct 3D analogue of standard doubled pawns, and no less a weakness here: neither pawn has a capture direction that reaches straight ahead, so neither can ever defend the other. What changes in three dimensions is the escape route. In flat chess the only way out is for one pawn to eventually advance past the other in rank. Here, either pawn can instead dissolve the structure by ascending a level — an axis 2D chess never offered. Raumcapa scores the configuration −20 cp to each pawn, a real but comparatively mild and removable penalty, applied because the weakness, however escapable, is still live until one pawn actually moves.
The Vertical Column: two pawns of the same color on the same file and rank but different levels (e.g., ♙︎Ac3 and ♙︎Bc3). It is tempting to read this as a support structure, with the lower pawn defending the upper — but that claim does not survive a check against this monograph’s own capture geometry, in Section II above: none of a pawn’s five capture directions is (L+1, f, r), so a pawn can never capture, and therefore never defend, the square directly above it. The two pawns are simply stacked, exactly as doubled pawns are stacked on a flat chess file, and the relationship is the same kind of liability: each blocks the other’s most natural advance. Raumcapa’s evaluation function, built independently from the same movement rules, agrees — it carries a level-doubled penalty (−20 cp to each pawn) for exactly this configuration, describing it in the same terms used here: a structure the less advanced of the two pawns can escape by advancing a rank instead, but a real weakness until it does.
Apex-Doubled Pawns: the permanent, severe case, occurring only on the topmost level a pawn can reach (Level E for White). A pawn on Level E that is also level-doubled with a pawn below it on Level D and blocked in its forward rank direction by a third own pawn ahead of it has no escape at all: the level advance is impossible because there is no level above E, and the rank advance is blocked by the pawn in front. Both routes out of the doubling are closed simultaneously, which never happens to the rank-doubled or ordinary level-doubled case below the apex. Raumcapa scores this configuration at −55 cp — roughly three times the ordinary doubling penalty — reflecting that the structure isn’t merely inconvenient but structurally immobile for as long as it persists.
The Diagonal Wedge: a pawn on Level B and a piece (typically a Unicorn) on Level C one square diagonally above it — e.g., pawn at Bc3 and Unicorn at Cc2. The most dynamic pawn-piece configuration in the opening, identified as ideal in Volume III.
The Isolated Level-C Pawn: a pawn alone on Level C with no adjacent same-color pawns on Level C and no piece support from Level B below. Weaker than an isolated pawn in flat chess, because it is exposed to three-dimensional attack from five directions.
The Raumschach pawn is simultaneously a builder (constructing pawn structures that support pieces) and a climber (seeking to ascend through the levels toward promotion). These roles are often in tension. Managing this tension — knowing when to build and when to climb — is one of the defining skills of Raumschach pawn play.
Philidor said that pawns are the soul of chess. In Raumschach, the pawn has acquired a second soul — an upward-striving ambition toward the sky of Level E that flat chess never gave it the space to dream of.
| Move type | (0,1,2) and permutations, ± |
| Max reach (center) | 24 squares from Cc3 |
| Color class | None — reaches all colors |
The Knight moves by the vector (0, 1, 2) — two of its three coordinates change, one by ±2 and one by ±1, while the third stays fixed. There are three ways to assign which coordinate changes by 0, ±1, and ±2, giving six base vectors; with signs, there are 24 distinct Knight moves in three dimensions (compared to 8 in flat chess). This is the largest pure-count expansion of the move set of any piece when going from 2D to 3D.
Knight at Cc3 = (3, 3, 3). All 24 possible destination squares: Fixing Level (ΔL=0): ΔF=±1,ΔR=±2 or ΔF=±2,ΔR=±1 (3,2,1)=Cb1 (3,4,1)=Cd1 (3,2,5)=Cb5 (3,4,5)=Cd5 (3,1,2)=Ca2 (3,5,2)=Ce2 (3,1,4)=Ca4 (3,5,4)=Ce4 Fixing File (ΔF=0): ΔL=±1,ΔR=±2 or ΔL=±2,ΔR=±1 (2,3,1)=Bc1 (4,3,1)=Dc1 (2,3,5)=Bc5 (4,3,5)=Dc5 (1,3,2)=Ac2 (5,3,2)=Ec2 (1,3,4)=Ac4 (5,3,4)=Ec4 Fixing Rank (ΔR=0): ΔL=±1,ΔF=±2 or ΔL=±2,ΔF=±1 (2,1,3)=Ba3 (4,1,3)=Da3 (2,5,3)=Be3 (4,5,3)=De3 (1,2,3)=Ab3 (5,2,3)=Eb3 (1,4,3)=Ad3 (5,4,3)=Ed3
From Cc3, the Knight reaches 24 squares distributed across all five levels simultaneously. This is the supreme quality of the Knight: it is the only piece that distributes its attacks uniformly across all five levels in a single move.
From a corner square (Aa1), the Knight reaches only 4 valid squares (versus 24 from the center). The contrast in mobility between center and corner (24:4 = 6:1) is steeper than for any other piece — Knights must be centralized aggressively to function. A Knight on Aa1 or Ae1 is doing almost nothing.
A Knight attacks two pieces on different levels in a single move — the three-dimensional fork unique to Raumschach.
Example: White Knight at Cc3=(3,3,3). Black Rook at Ac2=(1,3,2) — fixing file (ΔF=0), ΔL=−2, ΔR=−1: Cc3+(−2,0,−1)=(1,3,2)=Ac2 ✓. Black Queen at Ec4=(5,3,4) — ΔL=+2, ΔR=+1: (3,3,3)+(2,0,1)=(5,3,4)=Ec4 ✓. The Knight at Cc3 simultaneously attacks the Rook at Ac2 (2 levels below) and the Queen at Ec4 (2 levels above) — both on the c-file, separated by 4 levels and utterly invisible in any single-level diagram.
How to find it: Before every Knight move, check all five levels for pieces within the range of the Knight (0,1,2). The cross-level fork is found in the vertical scan, not the horizontal one.
The multi-target extension — three or more pieces: The Knight can fork not just two but three pieces simultaneously when its leap geometry places multiple attack vectors on occupied squares. The definitive example is the Cb5 Triple Fork — Pluto’s Attack (Opening 8, Volume III): Knight at Cb5 = (3,2,5) attacks the Black King at Ec5 via (+2,+1,0), the Black Rook at Ea5 via (+2,−1,0), and the Black Unicorn at Dd5 via (+1,+2,0) — three targets across two levels (E and D) and three different files. This is not a cross-level fork in the original sense (targets on the same file at different levels) but a full three-dimensional multi-fork: the Knight’s attack vectors vary simultaneously across the level, file, and rank coordinates. The pattern requires an Anchor piece (typically the Queen) to defend the forking square, making the opponent’s recapture economically unsound. Scan for it by counting how many enemy pieces fall within the Knight’s 24-square attack map from each candidate destination square — three hits is the maximum attainable and always decisive.
Confirmation from engineering: Raumcapa, by design, never searches an opponent’s reply — with exactly one carved-out exception. A dedicated scan checks every enemy Knight’s possible next leap and flags any landing square that would attack two or more own pieces at once, generating a defensive candidate before the fork ever lands. No other piece receives this treatment, for a precise reason: a slider’s future attack is visible the moment its line of sight opens, but a Knight’s is invisible until the leap is already made, so its threats can’t be read off the board the way every other piece’s can. That the engine’s sole exception to its own no-lookahead rule exists specifically for this pattern is an independent, code-level admission of what this section argues from geometry alone: the Knight fork is not just a tactic among many, but a structurally unique danger that ordinary reactive evaluation cannot see coming.
The Knight is one of only three pieces that can reach Level C on move 1 (alongside the Bishop and the Unicorn). From Ad1=(1,4,1), valid Knight moves include (+2,−1,0)→(3,3,1)=Cc1 — the Star Jump, closing the c-column with a Knight on move 1. The Knight also plays an important indirect opening role: moving to Ac3 or Ad3 on move 1 frees the Bishop or Unicorn behind it for immediate development on move 2.
The cross-level reach of the Knight gives it more endgame utility than its 2D counterpart: it can simultaneously pressure pieces on Levels A and E from a central Level C position. Two Knights together, from central positions, can create a web of threats that rivals the effectiveness of a Unicorn — though K+N+N vs. K, settled by exhaustive verification in Volume V, turns out to be a draw exactly as in flat chess; the wider 3D web of threats does not translate into a forced mate.
| Direction type | Edge — exactly 2 coords ±1 |
| Directional rays | 12 from any interior square |
| Color classes | Many — more complex than 2D |
The Bishop in Raumschach moves along edge-diagonal directions — the 12 directions where exactly two of the three coordinates change by ±1 simultaneously, while the third remains fixed. This is the natural generalization of the 2D diagonal: in two dimensions, a diagonal changes both file and rank by ±1; in three dimensions, we can fix any one of the three coordinates and vary the other two, giving 3 choices × 4 sign combinations = 12 directions total.
The 12 Bishop directions (exactly 2 coordinates change by ±1): Fixing Level (ΔL=0): (+0,+1,+1), (+0,+1,−1), (+0,−1,+1), (+0,−1,−1) → moves within a horizontal plane (same level) — same as standard 2D diagonals Fixing File (ΔF=0): (+1,+0,+1), (+1,+0,−1), (−1,+0,+1), (−1,+0,−1) → moves within a vertical rank-level slice Fixing Rank (ΔR=0): (+1,+1,+0), (+1,−1,+0), (−1,+1,+0), (−1,−1,+0) → moves within a vertical file-level slice
The Bishop has three types of diagonals: the familiar horizontal diagonals (within a level), and two types of vertical diagonals — one cutting through level-rank slices and one through level-file slices. These vertical diagonals are the unique contribution of the Bishop to Raumschach that it lacks in flat chess.
A Bishop on Ba1=(2,1,1) — its correct White starting square — can move along the level-file vertical diagonal (+1,+1,0) to Cb1=(3,2,1): this fixes rank (ΔR=0) ✓. This is the Bishop’s Flank, the game’s primary Bishop opening move. Continuing the ramp: from Cb1=(3,2,1), direction (+1,0,+1) → Db2=(4,2,2) ✓; from Db2=(4,2,2), direction (0,+1,+1) → Dc3=(4,3,3) ✓; from Dc3=(4,3,3), direction (+1,0,+1) → Ec4=(5,3,4) — delivering check to the Black King. The Bishop’s Flank ramp ♗︎Ba1–Cb1–Db2–Dc3–Ec4† is the game’s defining vertical-diagonal attacking sequence.
In flat chess, each Bishop is confined to one color — the “light-squared Bishop” never reaches a dark square. In Raumschach, the color-class structure is exactly analogous. Every Bishop move changes exactly two coordinates by ±1, so the sum L+f+r always changes by an even number (0 or ±2). The parity of (L+f+r) is therefore perfectly preserved across all 12 Bishop directions, dividing the 125 squares into two groups of 62 and 63.
Each Bishop is confined to exactly 62 squares — the half of the board sharing its (L+f+r) parity. This is the direct three-dimensional analogue of the flat chess color class, and here the analogy holds cleanly: the two White Bishops start at Ba1 (L+f+r=4, even) and Bd1 (L+f+r=7, odd) — opposite parities, exactly as the light-squared and dark-squared Bishops occupy opposite color complexes in flat chess. Together the two White Bishops therefore reach all 125 squares between them; no square is permanently safe from both. Whether this complementary coverage, together with the King, suffices to force spacemate against a lone defending King is a separate question from coverage alone: Volume V’s exhaustive verification settles it as a draw from almost every position, with forced spacemate available only from a narrow band already near a corner.
The Bishop’s vertical diagonals — the four directions that cross levels while staying on the same file or same rank — are its signature contribution to Raumschach that has no flat-chess parallel. These diagonals allow the Bishop to simultaneously threaten pieces on different levels along a straight “ramp” path through the three-dimensional board.
The most important vertical diagonals in opening theory are those that connect Level B to Level C to Level D — specifically the diagonal ♗︎Ba1–Cb1 (used in the Bishop’s Flank) and the attack path ♗︎Cb1–Db2–Dc3–Ec4 (the Bishop’s Flank attack diagonal). These diagonals allow the Bishop to make vertical progress — ascending toward the opponent’s home territory — while maintaining the long-range characteristics of a standard diagonal piece.
The Bishop travels along a vertical diagonal, “ramping” through levels and delivering an unexpected attack on a piece the opponent thought was safely out of range.
Characteristic position: White Bishop at Cc1=(3,3,1). Black Rook at Ec3=(5,3,3). The diagonal from Cc1 in direction (+1,0,+1): ♗︎Cc1–Dc2–Ec3. The Bishop ramps from Level C to Level D to Level E, attacking the Black Rook at Ec3. The opponent, thinking in horizontal terms, may not have noticed the Bishop at Cc1 can reach Ec3 — it looks like two levels and two ranks away, but along the vertical file-level diagonal it is a straight line of two steps. The Ramp Attack is the Bishop’s surprise weapon and the defining tactical motif of the Bishop’s Flank opening.
The Bishop is easily misassessed from two directions. Players arriving from flat chess may import the familiar valuation — Bishop roughly equal to Knight — and miss how dramatically the twelve-direction reach improves in three dimensions. Meanwhile the coverage table established in Maack’s 1919 analysis (1,752 total squares commanded across all starting positions in SIII5) places the Bishop above every other officer except the Queen. Both errors share the same root: failing to account for the vertical diagonals that are entirely new in three dimensions. The strategic error is placing Bishops only on horizontal diagonals within a single level — treating them as flat chess Bishops — and missing the ramp dimension entirely.
One clarification is worth stating plainly, because the natural assumption carried over from flat chess turns out to be correct here — unlike the case of the Unicorn pair (Monograph V). The two White Bishops, starting on opposite parities, complement each other exactly as in flat chess: together they cover the entire board, and no square is safe for the defending King simply by virtue of its color. Whether this complementary coverage is enough, on its own, to force spacemate against a bare King is settled in Volume V: only from a narrow band of positions already near a corner, not generally.
Raumcapa scores this exact configuration directly. Its evaluation function carries a dedicated Bishop Color Pair term: when a side’s two Bishops share the same color complex, the position is penalized 9 cp; when they cover opposite complexes instead — White’s actual starting situation, as established above — it is rewarded 18 cp, a 27 cp swing for a property that has nothing to do with material and everything to do with which half of the board the pair jointly covers. The engine did not need this volume’s analysis to find the reward; it falls directly out of the same parity argument made here, applied independently to the position both White Bishops actually start in.
| Direction type | Orthogonal — exactly 1 coord changes |
| Directional rays | 6 — ±Level, ±file, ±rank |
| Color class | None — reaches all squares |
The Rook moves along exactly one axis at a time — changing only the level, or only the file, or only the rank, by any number of steps in one direction. In three dimensions this gives six directional rays (±Level, ±file, ±rank) rather than four as in flat chess. The new pair of rays is the ±Level direction — the Rook can now move straight up or straight down through the levels, staying on the same file and rank. This is the new power of the Rook in Raumschach: the column.
The column — a set of five squares sharing the same file and rank but spanning all five levels (e.g., Ac3–Bc3–Cc3–Dc3–Ec3) — is the Rook’s uniquely three-dimensional territory. A Rook on any square of a column controls the entire column, from Level A to Level E. The column connects the two players’ home levels: a Rook ascending the c3-column from Ac3 to Ec3 pierces through all five levels, threatening pieces at every level along the way.
Raumcapa’s evaluation function agrees with the “more so” above by explicit calibration. Open and half-open lines earn every sliding piece a bonus scaled by a type factor reflecting how much that piece’s structure depends on open lines — and the Rook receives the highest factor of any piece, 1.0, ahead of the Unicorn’s 0.7, the Bishop’s 0.5, and the Queen’s 0.3. A Rook on a fully open line earns +15 cp; on a half-open line, +8 cp. The Queen’s factor is deliberately lower despite her far greater range — calibrated so that a centrally placed Queen on mostly open lines earns roughly the same total bonus as a centrally placed Rook, rather than dwarfing it simply by having more rays to be open. The Rook, in other words, is the piece the engine considers most defined by its open lines — exactly the dependency this section describes from first principles.
The Rook requires open lines to function, and in the opening its Rook on Level A, rank 1 is blocked in three directions. To activate, it needs one of these directions opened — typically the upward column, achieved by the King’s Corner Retreat which vacates the c1 square.
Before any Rook move in the opening or early middlegame, confirm at least one of the following: (1) the Rook is entering an open column; (2) the Rook is entering an open file on its current level; (3) the Rook is ascending the c1-column after the King has vacated it; (4) the Rook is delivering check or winning material immediately. A Rook moved where none of these conditions hold is a wasted tempo.
A Rook can reach any square on the board in at most two moves. This makes Rook-vs-passed-pawn races much more decisive than Knight or Bishop equivalents — the Rook can always intercept a pawn that has three or more moves to promotion, regardless of where the Rook starts. Only a pawn within two moves of promotion is truly beyond the reach of a Rook.
One specific Rook-and-pawn formation is valuable enough to name directly: the Tarrasch formation, a Rook standing behind its own passed pawn on a clear level or rank ray, supporting the pawn’s advance from the rear rather than escorting it from the side. Raumcapa rewards this placement with a standing +20 cp bonus whenever it detects a Rook with an unobstructed line to a passer in front of it — the three-dimensional engine independently arriving at the same formation Tarrasch identified in two dimensions over a century ago, because the underlying logic (the Rook gains the open line the pawn vacates as it advances) survives the addition of a third axis unchanged.
| Direction type | Triagonal — all 3 coords ±1 |
| Directional rays | 8 — space diagonals only |
| Color class size | 30 of 125 squares reachable |
The Unicorn is the only piece in the Raumschach series with no precursor in any standard or historical chess variant. It moves along the eight space diagonals — directions where all three coordinates simultaneously change by ±1. In three-dimensional geometry, these are the directions that point from the center of a cube toward each of its eight corners. They are called “triagonals” throughout this series: a form of motion that exists only because there are three coordinate axes, and vanishes the moment you remove any one of them.
Direction Mnemonic Main diagonal (from Aa1) (+1,+1,+1) → “The Ascender” Aa1–Bb2–Cc3–Dd4–Ee5 ★ THE MAIN TRIAGONAL (+1,+1,−1) → “The Climber” Aa5–Bb4–Cc3–Dd2–Ee1 (+1,−1,+1) → “The Farer” Ae1–Bd2–Cc3–Db4–Ea5 (+1,−1,−1) → “The Crosser” Ae5–Bd4–Cc3–Db2–Ea1 (−1,+1,+1) → “The Descender” Ee1–Dd2–Cc3–Bb4–Aa5 (−1,+1,−1) → “The Sinker” Ee5–Dd4–Cc3–Bb2–Aa1 (−1,−1,+1) → “The Returner” Ea1–Db2–Cc3–Bd4–Ae5 (−1,−1,−1) → “The Retreater” Ea5–Db4–Cc3–Bd2–Ae1 All eight triagonals pass through Cc3 — the absolute center. ★ The Main Triagonal connects corner Aa1 to corner Ee5.
The absolute center Cc3 lies on all eight triagonals simultaneously — the only square on the board with this property. However, Cc3 has parity (1,1,1), which falls outside every Unicorn’s color complex. No Unicorn can ever reach Cc3. A Queen placed there commands all eight space diagonals at once — but for each Unicorn the best attainable central squares depend on its color complex: the Bb1 Unicorn’s optimal squares are Cc2 and Cc4, both of parity (1,1,0); the Be1 Unicorn’s optimal squares are Bc3 and Dc3, both of parity (0,1,1).
Each Unicorn is permanently confined to exactly 30 of the 125 board squares. This is its most important and most consequential property. Each triagonal step changes (L+f+r) by an odd number (±1 or ±3), so parity of (L+f+r) alternates with every Unicorn step. The Unicorn therefore alternates between two parity classes: squares where (L+f+r) is even and squares where it is odd.
More precisely: the two White Unicorns start at Bb1=(2,2,1) [L+f+r=5, odd] and Be1=(2,5,1) [L+f+r=8, even]. They begin on squares of different fine-grained parity and therefore inhabit complementary, non-overlapping color complexes. The Bb1 Unicorn alternates between parity class (0,0,1) — 12 squares, L even, f even, r odd — and class (1,1,0) — 18 squares, L odd, f odd, r even. The Be1 Unicorn alternates between parity class (0,1,1) — 18 squares, L even, f odd, r odd — and class (1,0,0) — 12 squares, L odd, f even, r even. Together the two White Unicorns cover all four of these classes: 12+18+18+12 = 60 squares, with no overlap between their domains.
The 125 board squares divide into eight fine-grained parity classes (L%2, f%2, r%2). The two White Unicorns (Bb1 and Be1) inhabit complementary, non-overlapping domains:
Together: 60 squares. The remaining 65 squares — those of parity classes (0,1,0), (1,0,1), (1,1,1), and (0,0,0) — are beyond the reach of either White Unicorn. Note also that each White Unicorn shares its color complex with one Black Unicorn: the Black Unicorns start at Da5=(4,1,5) [class (0,1,1)] and Dd5=(4,4,5) [class (0,0,1)], meaning White’s Be1 Unicorn and Black’s Da5 Unicorn share a complex, as do White’s Bb1 Unicorn and Black’s Dd5 Unicorn. Unicorn-for-Unicorn captures between opponents on the same complex are therefore geometrically possible from the very first move.
Raumcapa’s evaluation function checks the complementary relationship established above directly, with no awareness of this volume’s reasoning. It carries a dedicated term comparing a side’s two Unicorns’ fine-grained parity classes: when they share the same class, the position is penalized 11 cp; when they cover complementary classes — White’s actual starting configuration — it is rewarded 22 cp, a 33 cp swing for a structural property invisible to a simple piece count. That White’s Unicorns happen to start complementary by the position’s original 1907 design, and that an engine reasoning from nothing but move geometry independently flags any departure from that arrangement as measurably worse, is a small, clean confirmation that the Dual Unicorn System is a fact about the board, not a habit of commentary.
The best attainable squares for each Unicorn depend on its color complex. For the Bb1 Unicorn: Cc2=(3,3,2) and Cc4=(3,3,4), both of parity (1,1,0), each giving 12 reachable squares — the maximum on a 5×5×5 board. For the Be1 Unicorn: Bc3=(2,3,3) and Dc3=(4,3,3), both of parity (0,1,1), likewise each giving 12 reachable squares. (Cc3 itself, the geometric center, is parity (1,1,1) and unreachable by any Unicorn.)
Compare this to a Unicorn in the corner (Aa1): from (1,1,1), only the direction (+1,+1,+1) is valid. The Aa1 Unicorn reaches only 4 squares (Bb2, Cc3, Dd4, Ee5) along its single available triagonal. The contrast between the corner Unicorn (4 reachable squares) and a best-placed central Unicorn — Cc2 or Cc4 for the Bb1 Unicorn, Bc3 or Dc3 for the Be1 Unicorn (12 reachable squares in each case) — is the starkest mobility differential of any piece — a 3:1 ratio. Keep Unicorns near the center. Always.
The Dead Unicorn — a Unicorn hemmed in by its own pawns on all eight triagonal directions — is the most catastrophic positional error unique to Raumschach. A Dead Unicorn literally cannot move: it is trapped, contributing nothing, sitting on a square of its color class while the battle rages elsewhere on the other 124 squares of the board.
Consider a White Unicorn hypothetically placed at Ba1=(2,1,1) — a corner square. Its eight triagonal directions from there:
(+1,+1,+1)→Cb2: if White pawn is on Cb2, the Unicorn cannot advance here. (−1,+1,+1)→invalid (Level 1, ΔL=−1 → Level 0). (+1,−1,+1)→invalid (file 1, ΔF=−1 → file 0). (+1,+1,−1)→invalid (rank 1, ΔR=−1 → rank 0). All other directions go off the board from a corner position.
From a corner square, the Unicorn has only ONE valid triagonal. If the only escape square is occupied by a friendly pawn, the Unicorn is completely trapped. The trap is created by a single pawn placement — a warning that even one thoughtless pawn move can imprison a Unicorn permanently.
General rule: Before advancing any pawn within triagonal distance of a Unicorn, verify that the Unicorn retains at least two open triagonal directions after the advance. Two open triagonals guarantee the Unicorn can always find an active path.
A Unicorn sacrifice is sound when it satisfies at least two of the following three conditions:
One empirical wrinkle from the 50,000-game Raumcapa self-play corpus deserves an honest mention here, because it cuts somewhat against this monograph’s own emphasis. Across the corpus, Unicorns move less often per piece than any other piece type — roughly 2.7 moves per Unicorn per game, against roughly 4.1 for Bishops, 4.0 for Rooks, and 3.1 for Knights. Two readings are available, and the corpus alone cannot decide between them. The favorable reading matches everything argued above: a Unicorn earns one strong square within its 30-cell domain and then sits there doing real work without needing to keep moving, exactly as a well-placed outpost piece should. The unfavorable reading is that Raumcapa’s depth-2 search simply does not look far enough to find the quieter Unicorn maneuvers — repositioning within the color complex, regrouping the pair — that a stronger player or a deeper-searching engine would, and defaults instead to whichever piece offers the most immediately calculable activity. Until a deeper-searching engine’s self-play is available for comparison, this stays an open question rather than a settled one, and it is recorded here for that reason rather than resolved.
Two Unicorns, exhaustively verified, cannot force spacemate against a lone King — the combined 60-square reach is too sparse to ever pin down every escape square of a cornered King at once. Two Bishops, by contrast, can force spacemate from a narrow band of already-cornered positions, though not from a generic one. Trading a Unicorn pair for a Bishop pair is therefore not the strategic error this volume once claimed; if anything, in the rare endgame where material has reduced to exactly this pairing, the Bishop pair is the one with any winning chances at all. The Dual Unicorn System’s real value lies in the middlegame coordination it provides (Volume IV), not in any bare-endgame mating power.
A single Unicorn in the endgame is at its best as an escort for a passed pawn — covering the pawn’s promotion square from a distance via the triagonal, so that enemy pieces cannot blockade without being captured.
The Unicorn is Raumschach’s gift to chess — the proof that three-dimensional chess is not merely flat chess with an extra layer, but a genuinely new game with genuinely new ideas. To play Raumschach and not understand the Unicorn is like playing flat chess and not understanding the bishop. You may survive, but you are playing in the dark.
| Directions | 26 — all non-zero unit vectors |
| Max squares | 26×4 — rays of up to 4 squares |
| Color class | None — all squares accessible |
The Queen combines the moves of every other piece except the Knight: Rook (6 orthogonal directions) + Bishop (12 edge-diagonal directions) + Unicorn (8 triagonal directions) = 26 directional rays. From the absolute center Cc3, the Queen at Cc3 threatens up to 52 squares in a single move — more than 40% of the entire board.
The 3D Queen is qualitatively different from any 2D queen: it can simultaneously threaten pieces on different levels via triagonals, pieces along ranks and files via orthogonals, and pieces along ramp-diagonals via its face-diagonal component. The 3D Queen is the only piece that can alone threaten pieces in all three spatial dimensions simultaneously.
A piece of such power comes with corresponding risks. The correct deployment of the Queen follows the principle established in Volume IV: the Queen should enter the position as the seventh step in the development sequence, after both Unicorns are coordinated, the c-column is secured, the King is in its corner, and the Rooks are activated. A Queen that enters earlier wastes time evading harassment rather than making threats. Exception: tactical openings that deploy the Queen in a specific defensive role (the Anchor Queen — see Role 4 below) may legitimately place the Queen on Level C on move 1 when a concrete, forcing material gain within three moves compensates entirely for the departure from the principled development order. Pluto’s Attack — the Cb5 Triple Fork (Opening 8, Volume III) is the defining example.
Raumcapa enforces a blunter version of the same principle as a hard gate rather than a guideline: the Queen may not leave her home square until at least one own Knight has already left its own home square. The gate is lifted only if applying it would leave the engine with no candidate moves at all — an escape valve for forced positions, not a loophole for ordinary play. A seven-step sequence is a refinement an engine without search depth cannot fully reconstruct; a one-condition gate is the coarsest version of the same instinct that survives without it. That coarse version still agrees with this section’s caution: not yet.
Role 1 — The Triagonal Battery Anchor. The Queen positions itself one triagonal step behind a Unicorn, forming a battery aimed at the enemy King. The Unicorn threatens directly; the Queen amplifies the threat and covers retreat squares. This is the Queen’s most powerful attacking role.
Role 2 — The Column Queen. The Queen occupies an open column (typically the c1-column or a-column) and applies vertical pressure. In this role, the Queen acts as a super-Rook, threatening along the column at full Queen range.
Role 3 — The Central Omnivore. The Queen occupies Cc3 (the absolute center) with adequate support and uses its 26-directional range to simultaneously threaten pieces on all five levels. This is the most ambitious role and requires careful preparation.
Role 4 — The Anchor Queen. The Queen is posted on a specific Level C square not to attack but to defend a forward Knight (or other piece) that occupies a tactical outpost, making the opponent’s capture of that piece unprofitable because the Queen recaptures with material gain. This role inverts the usual Queen logic: the Queen is not the threatening piece but the piece that makes a threat by another piece credible. The defining example is Pluto’s Attack, the Cb5 Triple Fork (Opening 8, Volume III): the Queen on Cb2 defends the Knight on Cb5, ensuring that if Black captures the forking Knight with the Queen (♕︎Dc5×Cb5), White replies ♕︎Cb2×Cb5 and wins the exchange. Remove the Queen from Cb2 and the Knight sacrifice is unsound; place it there and the combination is forced. The Anchor Queen is invisible to most players and is the hardest of the four roles to appreciate — but mastering it separates strong tactical players from weak ones.
In open positions, the Queen typically beats two Unicorns — her 26 directions overwhelm the Unicorns’ combined reach. In closed positions, the Unicorns may be superior: their triagonal directions cut through blocked positions more freely than the Queen’s orthogonal and face-diagonal rays.
| Reach (center) | 26 adjacent squares max |
| Reach (corner) | 7 minimum adjacent squares |
| Castling | None — no castling in Raumschach |
Of all seven piece types in Raumschach, the King is the one most changed — not in its movement rules, but in its strategic situation — by the transition from two to three dimensions. It begins on the central c-file, exposed to c-column threats, and must find its own safety through two carefully timed King moves — the Corner Retreat — while all other development proceeds simultaneously. The Corner Retreat is not optional: it is as fundamental to correct Raumschach play as castling is to flat chess.
Raumcapa’s development history makes the same point from an unexpected angle: it had to be built, debugged, and rebuilt to avoid getting this wrong. The engine’s general center-control term rewards any piece for occupying or influencing the central cube — correct for every piece that benefits from centralizing, which in the opening is all of them except one. As the engine’s own documentation puts it without qualification: “the King is the single piece in Raumschach that should explicitly not centralize in the opening, yet it benefits most from the [center-control] term by sheer geometric virtue of its 26 directions.” Left unguarded, the general-purpose centralization logic that correctly improves every other piece actively walks the King into the most dangerous structure on the board. The fix took two layers, applied redundantly on purpose: the King is excluded from the candidate-generating principles that would otherwise nominate centralizing moves for it, and a separate evaluation penalty (40 cp per level away from home, tripled inside the central cube) ensures that even if some future change reopened a path to a centralizing King move, the position score would still reject it. Two independent safeguards for one piece, where every other piece needs none, is itself a measurement of how differently the King must be treated once the board gains a third axis.
In the endgame, the King’s 26-direction reach makes it both harder to spacemate (more escape routes to cover) and more powerful as an attacking piece. The 26-direction King is the endgame’s dominant piece whenever the Queens have been exchanged — more mobile than any remaining Rook, Bishop, or Unicorn in the sheer number of squares it can reach in one step.
The 50,000-game Raumcapa self-play corpus bears this out with a specific wrinkle worth knowing. Spacemates concentrate heavily on Levels B and D (13,031 and 12,991 of 42,092 corpus spacemates) rather than on Level C itself (5,226) or the home corners A and E (5,292 and 5,552). The center is where the fight for control happens, but a King driven out of it is run down one level short of home — in the staging territory it must cross to reach safety — far more often than it is cornered in the center it just lost, or caught while still safely on its own level.
The endgame King’s optimal position is Level B or C, central file (b, c, or d) — providing maximum mobility while remaining at an altitude from which it can influence pieces on all five levels.
In Raumschach, king safety is never “solved.” The King in the corner is safer than the King on the c-column, but it is still exposed to attacks along the a-column (if the Rook moves away), to triagonal approaches from Bb2 or Cb1, and to long-range attacks from the opponent’s Queen along the face-diagonals of Level A.
The correct attitude to king safety in Raumschach is: constant, low-level vigilance rather than one-time resolution. In every position, spend one moment of the Position Checklist examining the three danger types (c-column exposure, triagonal approaches, level escalation) before proceeding to planning.
Recognizing that the Corner Retreat is correct is one thing; reliably choosing it is another, and Raumcapa’s own history separates the two. The engine generated the corner-approach move as a legal candidate from early on, but for some time gained nothing from playing it — a King sitting on the home level’s central file scored no worse, in isolation, than one a step closer to the corner, so the retreat was perpetually outbid by ordinary piece development with no urgency attached to it. The fix was a standing positional penalty tied directly to file distance from the corner: 50 cp lost on the central file, 25 cp one file out, zero at the corner file itself, applying only while the King remains on its home level. Each step of the retreat is therefore worth a concrete 25 cp, which is what finally made the engine choose the move it could already see. The lesson generalizes beyond software: knowing a principle and being incentivized to act on it immediately are different things, and a King’s safety, in particular, has to be made to matter now rather than eventually.
The analogous endgame principle to flat chess King activation is King altitude activation: bringing the King from Level A toward Level B or C, gaining elevation and therefore more mobility and influence over the pawn battles that typically decide Raumschach endgames. The ascending King that arrives at Level B, file c or d, rank 3 or 4 is in the ideal endgame position: commanding influence over all five levels and maximally supporting passed pawns wherever they may be on the board.
A second, more specific endgame technique is worth naming alongside altitude activation: interposing the King directly between the opponent’s King and one’s own passed pawn, so that the enemy King cannot approach the pawn without first passing the defending King. Raumcapa rewards exactly this geometry — an own King lying on a shortest path between the enemy King and an own passer — with +18 cp per qualifying pawn, a direct three-dimensional carry-over of a technique Capablanca used asymmetrically: a King escorting its own passer forward while simultaneously denying the opposing King any route to intervene.
The seven Raumschach pieces do not exist in isolation. They form an ecosystem — each piece complementing, competing with, and depending on the others.
| Pair | Relationship | Strategic Implication |
|---|---|---|
| Unicorn + Queen | The Triagonal Battery — the most lethal attacking combination. Queen amplifies the Unicorn’s triagonal threats. | Always seek this battery when attacking. The Queen behind a Unicorn on the main triagonal is the game’s supreme attacking formation. |
| Unicorn + Unicorn | Complementary color complexes — each Unicorn commands a different 30-square domain, together covering 60 squares with no overlap. Their non-overlapping territory makes them a genuinely coordinated attacking pair in the middlegame, but the combined reach is too sparse to deliver a bare-endgame mate (Volume V). | A real middlegame asset — the Dual Unicorn System still rewards coordination — but not, as once thought, the endgame’s strategic goal. Two Unicorns alone cannot force spacemate; do not expect this pairing to convert a bare endgame by itself. |
| Rook + Rook | Orthogonal coverage of two lines simultaneously; together they cover 10 of a face’s 25 squares, leaving 15 uncontrolled — the lawnmower intuition does not produce a reliable forced mate in 3D. | Exhaustively verified as a draw from almost every position (Volume V, Section VII): only 14,015 of 1,728,659 White-to-move positions are forced wins, all already near a corner. Strong restriction value in the middlegame; not a bare-endgame mating guarantee. |
| Bishop + Bishop | Complementary face-diagonals — together they cover 24 directional rays and, jointly, every square on the board. | Exhaustively verified to force spacemate only from a narrow band of positions already near a corner (Volume V) — not the general winning technique a 2D Bishop pair provides, but not the dead loss once attributed to it either. |
| Knight + Unicorn | Complementary leap-vs-slide character. The Knight attacks adjacent-level squares the Unicorn cannot reach (since Knights reach (0,1,2) vectors; Unicorns reach (1,1,1) vectors). | Together they cover a wider range of diagonal-type threats than either alone. A useful attacking pair in the middlegame. |
| Knight + Queen | The Anchor Battery — the Queen defends the Knight’s forward outpost, making a multi-fork economically viable. Introduced by Opening 8, Pluto’s Attack (the Cb5 Triple Fork): Queen on Cb2 defends Knight on Cb5, ensuring that Black’s capture of the forking Knight loses the exchange. | The Anchor Battery enables the Knight’s maximum tactical potential: a three-target fork that wins material regardless of whether the opponent captures the Knight (Queen recaptures with gain) or flees (Knight takes the most valuable unprotected target). Master this pairing to unlock Opening 8. |
| Rook + Unicorn | Orthogonal + triagonal coverage — together cover 14 directional rays with no overlap. The most efficient two-piece combination for area control. | Exhaustively verified as a complete draw against a lone King (Volume V, Section XIII): zero spacemate positions exist anywhere in the legal state space. Their joint 14-directional coverage is unmatched in the middlegame; in the bare endgame, the 12 uncovered face-diagonal directions always provide an escape route. |
| Pawn + Unicorn | The Diagonal Wedge — the pawn supports the Unicorn from one level below via the upward-forward capture direction. | The ideal opening structure. Pawn at Bc3 + Unicorn at Cc2 is the Raumschach equivalent of d4+Nf3 in flat chess. |
| King + Rook | The Rook can confine and harass a lone King, but cannot complete the job alone: exhaustive verification finds that no legal King/King/Rook position anywhere on the board ever has the defending King in check with zero legal replies (Volume V). | Settled: King and Rook alone is a theoretical draw against a lone King. The intuitive corner-mate construction fails because the only square close enough to cover all of a corner’s escape squares is itself illegally adjacent to the corner. |
Piece values in Raumschach are grounded in two independent sources: positional experience accumulated across this series, and Ferdinand Maack’s own piece value analysis, published in Raumschach: Einführung in die Spielpraxis (Hamburg, 1919). Maack’s §4c presents both an experiential ranking and a geometric coverage table — the total number of squares each piece commands when placed in turn on every square of SIII5. This coverage analysis is independent of rule variations or practical experience; it is a pure geometric fact about each piece.
| Piece | Total squares commanded in SIII5 | Maack’s experiential rank |
|---|---|---|
| Bishop | 1,752 | 3rd (Q and G above) |
| Rook | 1,500 | 4th |
| Knight | 1,452 | 2nd (Q and G above only) |
| Unicorn | 784 | 5th (P only below) |
This data shapes the hierarchy in three important ways. First, the Unicorn’s individual coverage — 784 squares, far below every other officer — reflects that the piece is permanently confined to one quarter of the board (30 squares per color complex). This is distinct from a pair premium: two Unicorns in their Dual Unicorn System, together commanding 60 squares, coordinate meaningfully in the middlegame even though (per Volume V’s exhaustive verification) that coordination does not extend to forcing spacemate in the bare endgame. The individual figure below is the value of a single Unicorn outside that system; the coordinated pair is worth more for positional reasons, though the earlier claim that this premium reflected forced-mate potential should be set aside. Second, the coverage table ranks the Bishop first among officers (1,752), well above the Rook (1,500), a real and important gap to keep in mind whenever the two are compared. Third, the Knight’s coverage (1,452) nearly matches the Rook’s, justifying much closer valuations between them than a naive read of “minor versus major” would suggest.
The Normal Form’s pawn structure shapes the hierarchy further. Maack’s valuations were calibrated in a game with five pawns per side on a single rank, with the extended three-direction pawn movement he called Pawn A. The Normal Form fields ten pawns per side on two ranks, with pawn movement restricted to two non-backward directions only. This considerably more closed opening depresses the Rook further — Maack already found it “cumbersome” in the sparser setup — while benefiting the Knight and Unicorn, whose leaps bypass the pawn mass freely. The resulting ordering is Q > B > N > R > U > P.
One clarification on the Bishop pair is worth stating plainly, since it confirms the natural assumption carried over from flat chess rather than overturning it: the two White Bishops start on opposite parities (Ba1 even, Bd1 odd) and do complement each other in coverage, together reaching the entire board exactly as the light- and dark-squared Bishops do in two dimensions. The Bishop pair’s value in Raumschach therefore rests on the same foundation it always has, plus the twelve-direction reach each Bishop gains from the third dimension.
Values are presented as a range to reflect the positional nature of value in three-dimensional chess. The Unicorn’s individual value is noted separately from its pair premium.
| Piece | Value (Pawns) | Range | Key Factors |
|---|---|---|---|
| Pawn | 1.0 | 0.5–2.5 | Passed Level-C and Level-D pawns worth much more; pawns near starting position on Levels A and B worth much less. In the Normal Form’s ten-pawn structure, creating a passer is harder to achieve — and therefore more decisive when it does appear. |
| Unicorn | 3.0 (individual) | 1.5–5.0 | Revised downward from 4.0. The coverage table is decisive: 784 total squares in SIII5, far below every other officer. A single Unicorn commands only its 30-square color complex. Pair value somewhat higher on coordination grounds (the Dual Unicorn System, Volume IV) — but the earlier claim that the pair is “worth 7.0 or more, enabling forced spacemate” should be retracted: Volume V’s exhaustive verification finds two Unicorns alone cannot force spacemate at all. Whatever premium the pair deserves comes from middlegame coordination, not endgame mating power. |
| Rook | 4.5 | 2.5–6.5 | Revised downward from 5.0. Maack’s “cumbersome” verdict is correct, and the Normal Form’s double pawn rank amplifies it: the Rook needs open lines that are genuinely rare in the closed opening. Value rises steeply once columns open — the endgame Rook on an open column approaches 6.5. The c1-column Rook remains the game’s most powerful activated piece, but the cost of activation must be counted against its raw power. |
| Knight | 5.0 | 2.0–7.0 | Revised upward from 3.5. Maack ranks the Knight second in his experiential scale (behind only Queen and Giraffe). Coverage (1,452) nearly matches the Rook. In the Normal Form, the Knight’s leap over the double pawn rank is a concrete practical advantage from move 1. Positional variance remains the steepest of any piece: 24 squares from the absolute center, 4 from a corner — centralize aggressively. |
| Bishop | 5.5 | 3.0–7.0 | Revised sharply upward from 3.5. Maack’s coverage table gives the Bishop the highest non-Queen score in SIII5 (1,752 squares). The twelve edge-diagonal directions — including the vertical ramp diagonals unique to three dimensions — give the Bishop extraordinary range when lines open. Note: the Bishop pair offers fully complementary coverage (Ba1 and Bd1 sit on opposite parities), exactly as in flat chess. Volume V’s exhaustive verification settles whether the pair alone can force spacemate: only from a narrow band of positions already near a corner, not generally — a real but limited endgame asset, well short of the flat-chess Bishop pair’s reliable technique. |
| Queen | 15.0 | 11.0–18.0 | Revised upward from 12.0. The 26-directional range — combining six orthogonal, twelve edge-diagonal, and eight triagonal rays — makes the 3D Queen qualitatively more dominant than either this volume’s first estimate or Volume IV’s preliminary figure of 11.0. The triagonal component alone gives the Queen forced-spacemate potential that no other single piece can match. |
| King | ∞ | — | Infinite (game-ending) value in check situations; roughly 4.0–5.0 practical endgame value as an active piece, given the 26-direction reach and its ability to escort passed pawns and participate in spacemate delivery across all five levels. |
The revision above rests on two sources: this volume’s own positional experience, and Maack’s 1919 geometric coverage analysis. A third has since become available — independent of both, and arrived at by a wholly different method. Raumcapa, the site’s Capablanca-modeled engine, assigns its own centipawn values, built from mobility-direction counts (rays or jump targets per piece) scaled by a per-direction weight, with the Unicorn’s color-binding applied as an explicit, deliberate exception to that scaling rather than a number arrived at by counting historical games or hand-tuning against play.
| Piece | Raumcapa (cp) | In Pawns | This Volume’s Figure |
|---|---|---|---|
| Pawn | 100 | 1.0 | 1.0 |
| Unicorn | 290 | 2.9 | 3.0 (individual) |
| Rook | 460 | 4.6 | 4.5 |
| Knight | 510 | 5.1 | 5.0 |
| Bishop | 560 | 5.6 | 5.5 |
| Queen | 1,620 | 16.2 | 15.0 |
The agreement is striking. Every figure lands within a tenth or two of this volume’s own estimate, and the ordering — Queen, Bishop, Knight, Rook, Unicorn, Pawn — is identical to the ordering derived from Maack’s 1919 coverage table. Three sources, separated by a century, built by entirely different means (lived practical judgment, hand-counted geometric coverage, and a mobility-weighted software heuristic), converge on essentially the same hierarchy. That convergence is meaningful: it is much harder for three independent methods to agree by coincidence than for any one of them to be wrong in the same direction three times.
The Unicorn’s story is the clearest case. Maack’s table flagged it as the weakest officer purely from counting squares — 784 of them, a fraction of any other piece’s reach. Raumcapa arrives at the identical conclusion through an unrelated mechanism: its evaluation function scales every piece’s value by direction count except the Unicorn’s, whose 290 cp is held fixed regardless of ray count specifically because color-binding caps its true reach at one-eighth of the board no matter how the count is measured. A 1919 geometer counting squares by hand and a 2026 engine designer writing an explicit exception in code arrived at the same insight about the same piece by two routes that share nothing but the board.
One caveat belongs here, stated plainly: Raumcapa’s figures are not a fourth authority so much as a third estimate. They derive from first-principles mobility scaling, not from texel tuning against recorded games — that tuning, against a dedicated self-play corpus, remains in progress. The convergence above stands as encouraging corroboration rather than final proof.
We have now studied all seven pieces of Raumschach individually — their geometries, their strengths, their weaknesses, their characteristic patterns, their interactions, and their values. What has this intimacy revealed?
Three observations stand out.
First: every piece is more positionally sensitive in three dimensions than in two. The mobility gradient — the difference between a piece’s power in the center versus the edge versus the corner — is steeper in every case. A Knight’s central mobility advantage (24 vs. 4) is more extreme in 3D than its 2D equivalent (8 vs. 2). A Unicorn’s central advantage (12 reachable squares from its optimal central squares — Cc2/Cc4 for the Bb1 Unicorn, Bc3/Dc3 for the Be1 Unicorn — vs. 4 from a corner) represents a 3:1 ratio. Centralization — already important in flat chess — is critical in Raumschach. A piece in the corner may be contributing almost nothing.
Second: the Unicorn is the piece around which the game’s unique character revolves. Remove the Unicorn from Raumschach and you have a reasonably interesting three-dimensional chess variant — but not a uniquely three-dimensional one. The Unicorn exists only because there are three dimensions. Its triagonal movement has no meaning in two dimensions. Its color-class structure, its endgame sufficiency theorems, its role in the Dual Unicorn System, its Dead Unicorn pathology — all of these are genuinely three-dimensional phenomena. Maack’s genius was inventing this piece: the piece that makes Raumschach not “chess with layers” but chess transformed by space.
Third: the pieces form a community more than a collection. The Pawn and Unicorn collaborate in the Diagonal Wedge. The Queen and Unicorn form the Triagonal Battery. The King and Rook deliver spacemate in the corner. The two Unicorns reinforce each other across complementary 30-square color complexes, together covering 60 squares. The Bishop’s vertical diagonals complement the Unicorn’s triagonals. None of these relationships exists in flat chess; all of them emerge from the three-dimensional geometry that Maack gave the game in 1907. The pieces were there, in modified form, in flat chess — but the space between them changed them, and changed how they relate to each other, and created a community that is richer than the sum of its members.
This is the first volume of the first theoretical series ever written for Raumschach. Six volumes, from opening principles to endgame technique to strategic axioms to annotated games to the intimate geometry of each individual piece. It took 119 years to begin. May it take far less time to continue. And may Maack himself remain a guide: his 1919 coverage analysis, dormant for over a century, anchors the value hierarchy in Section IX above. That a piece of software built in 2026, knowing nothing of Maack’s book, should arrive independently at his same hierarchy is its own kind of confirmation: some truths about this board were always there to be found, by whoever looked carefully enough, in whatever century they happened to be looking.
Ferdinand Maack believed that chess should reflect the three-dimensional reality of the world. Having now lived inside his game long enough to write these pages, the conviction is impossible to resist: he was right. The flat board is a beautiful abstraction. The cubic board is something closer to the truth.