Symbols on a page hide something obvious: 3x + 2x is just five things of the same kind gathered together. The tiles in this widget make that obvious by giving each term a shape. Long rectangle = one x. Small square = one 1. Negatives = the same shape in a muted colour. Sliders add or remove tiles; the canvas draws them in neat rows; the readout tells you the combined expression. Drag, change counts, and you will see why combining like terms is a physical act — stacking tiles that are literally the same shape — and why you cannot combine xs with 1s any more than you can combine forks with spoons.
The physical model
Algebra tiles were invented for classrooms in the 1970s to fix a specific confusion: students knew how to write 3x + 2x = 5x but not why it worked. Tiles make the rule visible.
- A long tile has length x and width 1. Its area is x \cdot 1 = x. One long tile represents one copy of the variable x.
- A small tile is a 1 \times 1 square. Its area is 1. One small tile represents the constant 1.
- Negatives are the same two shapes in a faded, outlined colour. Add one long-positive tile and one long-negative tile and they cancel — their areas are +x and -x, which sum to zero.
Because the two shapes are different, you can see that you cannot merge them. A long tile is not made of small tiles — it would take x of them, and x is a variable, not a fixed number. The shapes themselves refuse to combine.
Widget — drag and count
Start with the defaults — 3 long tiles and 2 small tiles. The readout shows 3x + 2. Push the +x slider to 5; you now have five long tiles and the readout says 5x + 2. Pull the -x slider up to 2; two negative long tiles appear beside the five positives. Press Regroup and they pair up and cancel, leaving 3x + 2 — the same expression you started with, because 5x - 2x = 3x.
Combining like terms as a shape game
The central rule of simplification is: only tiles with the same shape can merge. Try it. Set +x = 3, +1 = 2, -x = 0, -1 = 0. The tray shows three long tiles and two small tiles. Can you line the small tiles up flush against the long ones to form a bigger block? No — a row of two unit squares is 2 units long, but each x-tile is x units long, and x is a variable. The shapes don't fit.
Now set +x = 3 and +x = 5 separately… actually, you can't, because both are the same shape in the widget. But that is exactly the point: in 3x + 2x, both groups are long tiles, so the widget naturally merges them into 5x. The answer falls out of the geometry.
The negative tiles show why subtraction works. 4x - x is four positive longs and one negative long. Press Regroup and one positive pairs with the negative — their signed areas +x and -x sum to zero — leaving three longs. The widget reads 3x. No algebra was needed; the shapes cancelled.
Worked example — 3x + 2 + 2x + 5
The canonical exercise, tile-by-tile.
- Lay out 3 long tiles. Readout: 3x.
- Add 2 small tiles. Readout: 3x + 2.
- Add 2 more long tiles. Readout: 5x + 2.
- Add 5 more small tiles. Readout: 5x + 7.
You never performed an operation. You just placed tiles and read off what was on the tray. Each long tile kept its identity as an x; each small tile kept its identity as a 1. The final answer 5x + 7 is what the tray is. This is what "combining like terms" means in the abstract: gathering all tokens of the same type and counting them, separately for each type.
Try the same sequence with negatives. Lay 4 positive longs. Add 3 negative longs. Press Regroup. Three positive-negative pairs cancel, leaving one long — the readout shows x. That is the tile-level view of 4x - 3x = x.
Why the shapes themselves encode the algebra
The genius of the tile model is that the rules of algebra come out of geometry, not from a teacher saying "this is the rule, memorise it." Three rules, three geometric facts.
- Commutativity of addition — 3x + 2 = 2 + 3x. Rearrange the tiles on the tray in any order; the collection is the same. Addition of expressions is commutative because addition of tile-counts is commutative.
- Combining like terms — 3x + 2x = 5x. Three long tiles next to two long tiles give a group of five long tiles. The coefficient is literally the count of that shape on the tray.
- Cancellation of opposites — x + (-x) = 0. A positive long tile and a negative long tile have signed areas +x and -x. Together they occupy one long shape's worth of tray, but the signed total is zero, so you can remove the pair without changing what the tray represents.
Without tiles, those three facts look like three separate rules you memorise. With tiles, they are one rule — count tokens of each type — repeated in three situations.
What the model hides — and when to leave it
Tiles are great for linear expressions nx + m. They extend to quadratics (square tiles of area x^2) and to multiplication by rectangle area (the area-model figure in algebraic expressions). But they do not scale past degree 2 — there is no fourth-dimension tile for x^3 — and they become clumsy for expressions in two variables or with fractional coefficients.
There are other limits. The model works only for integer coefficients because you cannot place half a tile. It hides the distributive law when a coefficient multiplies a bracket: 2(x + 3) has a clean tile reading (two copies of "one long tile and three small tiles"), but a fraction like \tfrac{1}{2}(x + 3) does not. The model also suggests that combining like terms is always visual, but in expressions like (x + 1)^2 - (x - 1)^2 the cancellations are algebraic, not geometric; you expand, collect, and cancel using the distributive law rather than pushing tiles around.
The right moment to leave the tiles behind is when the answer is obvious without them. If you already see 3x + 2x = 5x without moving a single tile, the model has done its job. The tiles were training wheels for a rule ("gather same-shaped tiles") that you now apply in your head. For the full algebraic framework — terms, coefficients, degree, and the distributive law that makes all of this work — return to algebraic expressions.