How to Master Shikaku
TLDR: Shikaku is an area-partition logic puzzle - drag rectangles so every number ends up inside one rectangle whose area matches it exactly, with no gaps or overlaps. Start with the most constrained clues (prime numbers, corner cells), list their valid factor pairs, cascade the forced placements outward, and backtrack calmly when contradictions appear.
How the Game Works
Shikaku’s rule is simple: partition the entire grid into non-overlapping rectangles. Every rectangle must contain exactly one number, and that number must equal the rectangle’s area. When every cell belongs to a rectangle and every rectangle matches its clue, the puzzle is solved.
You drag to draw rectangles on the grid. Every cell must belong to exactly one rectangle. Every clue number must sit inside a rectangle of matching area. No exceptions - it is all or nothing.
The invisible constraint system is what makes Shikaku compelling. A “12” clue could become a 1x12, 2x6, 3x4, 4x3, 6x2, or 12x1 rectangle. The other clues and the grid’s geometry force you to pick the right one. This interplay between local flexibility and global rigidity is what separates casual solving from mastery.
Before drawing anything. Add up all the clue numbers on the board. That sum must equal the total cell count. If it does not, you have misread a number. This 5-second check prevents wasted effort.
The Core Skill: Factor Sense and Spatial Decomposition
Shikaku trains two connected abilities.
Factor sense is multiplication intuition. When you see a “20”, you stop seeing a number and start seeing 1x20, 2x10, 4x5, 5x4, 10x2, 20x1. You learn which factor pairs are common, which are rare, and which are geometrically impossible in a given grid. After a few sessions, you enumerate factor pairs automatically and instantly rule out dimensions that cannot fit the remaining space.
Spatial decomposition is the ability to mentally cut a space into regions and verify that each region satisfies its constraints. You hold multiple candidate partitions in mind simultaneously and test them for consistency. This is the same reasoning architects, urban planners, and circuit designers use daily.
Together these skills mean you are not guessing. You are systematically collapsing possibility space by reasoning about constraints and their interactions.
Fundamentals First
Every cell must be used. If you finish and any cell is empty, the solution is wrong. This single rule eliminates “solutions” that look right locally but leave orphan cells.
Learn factor pairs for 1 through 20. You do not need to memorise them, but five minutes listing them builds muscle memory. A “6” is 1x6, 2x3, 3x2, 6x1. A “15” is 1x15, 3x5, 5x3, 15x1. Knowing these automatically saves you from entering dead ends.
Always start with the most constrained clues. Prime numbers (2, 3, 5, 7, 11, 13) have only two factor pairs: 1xN and Nx1. A “7” anywhere on the board is either a 1x7 or a 7x1 rectangle - nothing else. These tight clues often force their shapes immediately, which then constrains neighbouring clues.
Primes unlock the board. Identify all prime number clues first. Their rectangles have only two possible orientations. Placing them creates a skeleton of fixed regions that composites (12, 15, 20) must fill around.
Core Strategy: Constraints Cascade Inward
The winning approach reverses the intuitive one. Instead of asking “where could this number go?”, ask “what definitely does not fit here?”
Constraint cascade. Find the most restricted clue - fewest valid factor pairs, nearest to edges or corners, most surrounded by already-placed rectangles. Force its rectangle into the only shape that fits. That shrinks available space for adjacent clues. Repeat until the solution is forced. You are not placing rectangles; you are eliminating impossibilities until one option remains.
Edge anchor. Clues near corners or edges have fewer valid rectangle orientations because the grid boundary cuts off options. A “4” in a corner with only 3 rows below it cannot form a 4x2 going down - you can eliminate that orientation immediately. Find these anchors and lock them in first. Their neighbours then become easier.
Work systematically from high-constraint to low-constraint clues. Prime numbers and small clues (2, 3, 5) almost always have forced shapes. Composite numbers with many factors (12, 18, 20, 24) have more options and should come later. Solving the primes first creates a framework the composites must fit inside.
Common Mistakes
Leaving orphaned cells. You place five rectangles perfectly, only to find one cell left over that matches no remaining clue. This happens because you did not plan far enough ahead. Before finalising a rectangle, ask: “If I place this here, can the remaining empty cells be partitioned exactly by the remaining clues?” Their sum must equal the remaining empty cell count.
Orphan check. Before placing a rectangle, tally the remaining clue numbers. Do they sum to the remaining empty cells? If not, your current placement is wrong. This check takes 10 seconds and prevents the most common class of dead ends.
Tunnel vision on one clue. You become convinced a “12” must be 3x4 and force it there. Only later does a contradiction emerge elsewhere. Hold hypotheses lightly. For any clue with three or more factor pairs, trace two moves forward from each option. The one that creates the most forced placements is usually correct.
Hypothesis test. For clues with 3 or more factor pairs, list all valid dimensions that fit the grid’s geometry. For each option, trace forward two placements. Which option leaves the neighbouring clues most constrained? Choose that one. The best move is the one that most restricts future choices, not the one that feels spatially natural.
Ignoring geometry. A “6” could be 1x6 or 2x3, but if you are near the board edge with only 2 rows available, a 1x6 going vertically is impossible. Always confirm a rectangle’s dimensions actually fit in the remaining space before treating it as a valid option.
Geometry check. Before placing a rectangle, verify it does not extend beyond grid boundaries and does not overlap any already-placed rectangle. A valid factor pair that cannot physically fit in the remaining space is not a valid option for this puzzle.
Solving greedily without reading the whole board. You solve the top-left corner perfectly and then realise the remaining space cannot be partitioned by the remaining clues. Spend 10 seconds scanning the entire board before drawing the first rectangle. Understand the clue distribution. Notice any unusual concentrations of large or small numbers.
Advanced Tactics
Forced propagation. Place one carefully chosen rectangle and trace every consequence. A single well-placed rectangle often cascades into half the puzzle solving itself, especially near edges where neighbours have limited options.
Contradiction engine. When stuck, pick an uncertain clue and assume one of its factor pairs is correct. Solve forward until you either finish the puzzle or hit an impossible state. If you hit a contradiction, that factor pair is wrong - try the next one. This method is slow but guaranteed to work on any valid puzzle.
Symmetry shortcut. Some puzzles have vertical or horizontal symmetry. If you solve one half perfectly, the other half often mirrors it. Spot this early and you can reduce total work significantly.
Mastery signal. You are improving when you can scan a board, identify the three or four most constrained clues, and mentally trace their forced placements without drawing anything. When you can predict the skeleton of a solution before committing a single rectangle, you are thinking at expert level.
7-Day Practice Routine
Days 1-2: Solve three easy boards. Start only with 2s, 3s, and 5s. Time yourself. Target 3-5 minutes per easy board.
Days 3-4: Solve three medium boards. Before drawing anything, spend 30 seconds identifying the top three most constrained clues. Solve only those first. Then assess what is forced by those placements.
Days 5-6: Solve two medium and one hard board. For every clue with multiple factor pairs, explicitly list the valid options and test the most constraining one first.
Day 7: Solve one easy, one medium, and one hard board in sequence. Total time for all three should be under 20 minutes. You are training pattern recognition, not just logic.
Repeat the cycle weekly, gradually pushing harder boards into the medium and easy slots.
Time yourself per board. An easy board should take 3-5 minutes, medium 8-12 minutes, hard 15-20 minutes at first. Track your times over three weeks. Faster times indicate that factor sense and spatial decomposition are becoming automatic - the clearest signal that the skill is consolidating.
Practice payoff. After three weeks of consistent play, clues will start resolving before you consciously enumerate their factor pairs. Factor sense and spatial decomposition are becoming automatic rather than effortful - which is exactly the point.
Shikaku
Divide the grid into rectangles · each one holds a single number equal to its area. A clean area-partition logic puzzle
Play nowWorks on any device.
