In short
The scalar triple product of three vectors \vec{a}, \vec{b}, \vec{c} is the number \vec{a} \cdot (\vec{b} \times \vec{c}), written [\vec{a}\;\vec{b}\;\vec{c}]. Geometrically, its absolute value is the volume of the parallelepiped formed by the three vectors. When this number is zero, the three vectors are coplanar — they all lie in one flat plane. Algebraically, it equals the determinant of the 3 \times 3 matrix whose rows are the components of the three vectors.
Take three sticks of different lengths and different directions. Tape their tails together at one corner. Now imagine building a box — not a cube, but a slanted box — by extending each stick into a full edge and filling in the six faces with parallelograms. The shape you get is called a parallelepiped (say it like "par-uh-lel-eh-PIE-ped"). It is the three-dimensional cousin of a parallelogram.
Here is the question: what is the volume of this box?
With a cube, volume is just side cubed. With a rectangular box, it is length times width times height. But a parallelepiped is slanted — its edges are not perpendicular — so none of those simple formulas apply. You need something that accounts for the angles between the edges, not just their lengths.
The answer turns out to be a single operation on the three edge vectors. It combines a cross product and a dot product in one step, producing a single number — a scalar. That number is called the scalar triple product, and it gives you the volume directly.
Building the formula
Suppose the three edges meeting at one corner of the parallelepiped are the vectors \vec{a}, \vec{b}, and \vec{c}.
Start with two of them, say \vec{b} and \vec{c}. Their cross product \vec{b} \times \vec{c} is a vector that is perpendicular to the parallelogram formed by \vec{b} and \vec{c}, and whose magnitude equals the area of that parallelogram:
where \theta is the angle between \vec{b} and \vec{c}.
Now you have the base of the parallelepiped (the parallelogram with area |\vec{b} \times \vec{c}|) and a direction perpendicular to it (\vec{b} \times \vec{c} points straight "up" from the base). The volume of the box is the base area times the perpendicular height — the component of \vec{a} in the direction of \vec{b} \times \vec{c}.
That perpendicular component is exactly what a dot product extracts. The projection of \vec{a} onto the direction of \vec{b} \times \vec{c} is
So the volume is
The magnitude cancels, and you are left with just the dot product of \vec{a} with the cross product of \vec{b} and \vec{c}. That product can be negative (if \vec{a} points "below" the base rather than "above" it), so the volume is the absolute value:
The definition
Scalar triple product
For three vectors \vec{a}, \vec{b}, \vec{c} in \mathbb{R}^3, the scalar triple product is
If \vec{a} = a_1\hat{i} + a_2\hat{j} + a_3\hat{k}, \vec{b} = b_1\hat{i} + b_2\hat{j} + b_3\hat{k}, \vec{c} = c_1\hat{i} + c_2\hat{j} + c_3\hat{k}, then
The volume of the parallelepiped formed by \vec{a}, \vec{b}, \vec{c} is |[\vec{a}\;\vec{b}\;\vec{c}]|.
Why it equals a determinant
This is worth seeing. Write out \vec{b} \times \vec{c} in component form:
Now take the dot product with \vec{a} = a_1\hat{i} + a_2\hat{j} + a_3\hat{k}:
Look at the right side. That is exactly the cofactor expansion of a 3 \times 3 determinant along the first row — with a_1, a_2, a_3 as the first row, and the 2 \times 2 subdeterminants formed from the remaining rows. So
The scalar triple product is a determinant. This connection is one of the reasons determinants matter in geometry — a 3 \times 3 determinant is, quite literally, a volume.
Properties
The scalar triple product has several clean properties. Each follows from determinant properties.
1. Cyclic permutation preserves the value.
Cycling the three vectors — moving the first to the end and shifting the rest forward — does not change the determinant. (In determinant language, two adjacent row-swaps bring you back to the same sign: swapping rows once flips the sign, swapping again flips it back.)
Equivalently, the "dot" and the "cross" can be interchanged:
This is sometimes written as: "the dot and cross are interchangeable in a scalar triple product." Geometrically, the volume of the parallelepiped does not depend on which face you call the base.
2. Swapping any two vectors flips the sign.
Swapping two rows of a determinant flips its sign. So [\vec{a}\;\vec{b}\;\vec{c}] = -[\vec{b}\;\vec{a}\;\vec{c}]. The magnitude (and hence the volume) stays the same, but the sign changes. The sign tracks the orientation — whether the triple (\vec{a}, \vec{b}, \vec{c}) forms a right-handed or left-handed system.
3. Scalar factor pulls out.
Scaling one vector scales the volume proportionally. This follows from the determinant property that multiplying one row by \lambda multiplies the determinant by \lambda.
4. Linearity in each argument.
The scalar triple product distributes over vector addition in any one slot. This is the multilinearity of the determinant.
5. If any two vectors are equal, the product is zero.
A determinant with two equal rows is zero. Geometrically, if two of your edge vectors are the same, the "parallelepiped" is flat — it has no volume.
The coplanarity test
This is the property that makes the scalar triple product indispensable. Three vectors \vec{a}, \vec{b}, \vec{c} are coplanar — they all lie in one flat plane — if and only if the parallelepiped they form has zero volume. And that happens exactly when
Turn this around: four points A, B, C, D are coplanar if and only if the three vectors \overrightarrow{AB}, \overrightarrow{AC}, \overrightarrow{AD} have scalar triple product zero.
Why does this work? If \vec{a}, \vec{b}, \vec{c} are coplanar, then \vec{a} lies in the plane spanned by \vec{b} and \vec{c}. But \vec{b} \times \vec{c} is perpendicular to that plane. So \vec{a} is perpendicular to \vec{b} \times \vec{c}, which means their dot product is zero. Conversely, if \vec{a} \cdot (\vec{b} \times \vec{c}) = 0, then \vec{a} has no component along \vec{b} \times \vec{c} — it lies entirely in the \vec{b}-\vec{c} plane.
Coplanarity of four points
To check whether four points A(x_1, y_1, z_1), B(x_2, y_2, z_2), C(x_3, y_3, z_3), D(x_4, y_4, z_4) are coplanar, compute:
If this determinant is zero, the four points lie in one plane.
Volume of a tetrahedron
A parallelepiped and a tetrahedron are related. A tetrahedron with vertices A, B, C, D occupies exactly \frac{1}{6} of the parallelepiped formed by the vectors \overrightarrow{AB}, \overrightarrow{AC}, \overrightarrow{AD}. (The factor \frac{1}{6} plays the same role as the \frac{1}{2} in the triangle area formula — a triangle is half its parallelogram, and a tetrahedron is one-sixth of its parallelepiped.)
This formula is used constantly in 3D geometry problems.
Computing one from start to finish
Example 1: Volume of a parallelepiped
Find the volume of the parallelepiped whose three co-terminal edges are \vec{a} = 2\hat{i} - 3\hat{j} + 4\hat{k}, \vec{b} = \hat{i} + 2\hat{j} - \hat{k}, \vec{c} = 3\hat{i} - \hat{j} + 2\hat{k}.
Step 1. Write the scalar triple product as a determinant.
Why: the three vectors form the rows, in order. The scalar triple product is this determinant.
Step 2. Expand along the first row.
Why: cofactor expansion — each element of the first row multiplies the 2 \times 2 determinant left after deleting its row and column, with alternating signs +, -, +.
Step 3. Compute each 2 \times 2 determinant.
Why: a 2 \times 2 determinant is ad - bc. Careful with the signs — 1 \times 2 - (-1) \times 3 = 2 + 3 = 5.
Step 4. Combine.
Why: the sign of -3 in the first row was handled by the cofactor sign pattern. The entry is -3, the cofactor sign is -, so the contribution is -(-3) \times 5 = +15.
Step 5. The volume is the absolute value.
Result: The parallelepiped has volume 7 cubic units.
The negative sign in the scalar triple product has a geometric meaning: it tells you about orientation. A positive value means (\vec{a}, \vec{b}, \vec{c}) form a right-handed system (like the standard \hat{i}, \hat{j}, \hat{k}); a negative value means they form a left-handed system. The volume itself is always the absolute value.
Example 2: Testing four points for coplanarity
Are the four points A(1, 2, 3), B(4, 0, 1), C(2, 1, 5), D(7, -1, 3) coplanar?
Step 1. Compute the three vectors from A to the other points.
Why: four points are coplanar exactly when the three vectors connecting one of them to the other three have scalar triple product zero.
Step 2. Set up the determinant.
Why: the three vectors become the three rows of a 3 \times 3 determinant.
Step 3. Expand along the third row (the zero makes it efficient).
Why: expanding along the row with a zero entry saves one minor computation. The cofactor signs for the third row are +, -, +.
Step 4. Evaluate.
Result: The scalar triple product is -12 \neq 0, so the four points are not coplanar. They span a tetrahedron of volume \frac{1}{6}|{-12}| = 2 cubic units.
If the determinant had come out to zero, you would have concluded that the four points are coplanar. The scalar triple product is a single-number test: zero means coplanar, nonzero means not coplanar.
Common confusions
-
"The scalar triple product is a vector." No — despite involving two vector operations (a cross product and a dot product), the final result is a scalar (a single number). The cross product produces a vector, and the dot product collapses it into a number. That is why it is called the scalar triple product.
-
"The order of the vectors does not matter." It does. Swapping two vectors flips the sign. The volume (the absolute value) is the same, but the sign changes. Cyclic permutations preserve the sign; non-cyclic ones flip it.
-
"\vec{a} \cdot (\vec{b} \times \vec{c}) and (\vec{a} \cdot \vec{b}) \times \vec{c} are the same thing." The second expression makes no sense. \vec{a} \cdot \vec{b} is a scalar, and you cannot take the cross product of a scalar with a vector. The parentheses matter: the cross product must be computed first, then the dot product.
-
"If the scalar triple product is zero, the vectors are zero." Not at all. It means the three vectors are coplanar — they lie in one plane. All three can be nonzero, long, and interesting. They just happen to be flat, with no "depth" perpendicular to their common plane.
-
"A tetrahedron has volume [\vec{a}\;\vec{b}\;\vec{c}]." Close, but off by a factor of 6. A tetrahedron is \frac{1}{6} of the parallelepiped. This is the three-dimensional version of the mistake of confusing a parallelogram's area with a triangle's area.
Going deeper
If you came here to learn what the scalar triple product is, how to compute it, and what zero means, you have it — you can stop here. The sections below explore how the scalar triple product connects to orientation, how it behaves under coordinate transformations, and its role in the volume element of integration.
Orientation and the sign
The sign of [\vec{a}\;\vec{b}\;\vec{c}] carries geometric information. If you curl the fingers of your right hand from \vec{b} toward \vec{c} and your thumb points roughly along \vec{a}, the triple product is positive — the system is right-handed. If your thumb points opposite to \vec{a}, the triple product is negative — the system is left-handed.
The standard basis (\hat{i}, \hat{j}, \hat{k}) is right-handed, and
Any set of three vectors that forms a right-handed system has a positive scalar triple product. This idea — that a determinant encodes orientation — shows up again in the Jacobian of coordinate transformations, where the sign tells you whether the transformation preserves or reverses orientation.
Connection to the cross product magnitude
There is a useful identity relating the scalar triple product to magnitudes:
This is a three-dimensional analogue of the identity \cos^2\theta + \sin^2\theta = 1, applied to the angle between \vec{a} and \vec{b} \times \vec{c}. The scalar triple product captures the \cos\theta part (the projection), while the vector triple product (which you will meet in the next article) captures the \sin\theta part (the rejection).
The volume element
In multivariable calculus, when you change from Cartesian to curvilinear coordinates (spherical, cylindrical, etc.), the volume element dV picks up a factor that is exactly a scalar triple product of the new basis vectors — the Jacobian determinant. The same idea of "three vectors spanning a tiny parallelepiped" reappears in a more sophisticated form.
Where this leads next
The scalar triple product sits at a junction between several important ideas.
- Vector Triple Product — what happens when you take \vec{a} \times (\vec{b} \times \vec{c}), a cross product of a cross product, and get a vector instead of a scalar.
- Cross Product — the operation that gives the base area, which the scalar triple product extends into a volume.
- Dot Product — the operation that extracts the height, which the scalar triple product combines with the cross product.
- Determinants — Introduction — the algebraic engine behind the scalar triple product, and the reason its properties are so clean.
- Collinearity and Coplanarity — the broader setting for the coplanarity test, including the linear-dependence perspective.
- Reciprocal System of Vectors — a construction that uses the scalar triple product in every one of its formulas.