[proofplan] We obtain the Divergence Theorem as a special case of the [Generalised Stokes' Theorem](/theorems/3555) applied to a single $2$-form on $\mathbb{R}^3$. The strategy has three ingredients: (i) encode the vector field $F$ as the **flux $2$-form** $\omega = \iota_F\, \Omega$, where $\Omega = dx_1 \wedge dx_2 \wedge dx_3$ is the standard volume form; (ii) compute $d\omega = (\nabla \cdot F)\, \Omega$, identifying the divergence with the [exterior derivative](/theorems/1525) of $\omega$; (iii) check that the pullback of $\omega$ to the oriented boundary $\partial M$ coincides pointwise with the surface flux density $(F \cdot n)\, dS$. [Stokes' Theorem](/theorems/1530) $\int_M d\omega = \int_{\partial M} \omega$ then becomes the asserted equality once we translate form integrals into measure integrals via the standard volume and surface forms. [/proofplan] [step:Introduce the flux $2$-form $\omega = \iota_F\, \Omega$ associated to $F$] Let $\Omega \in \Omega^3(\mathbb{R}^3)$ denote the standard volume form \begin{align*} \Omega = dx_1 \wedge dx_2 \wedge dx_3, \end{align*} which orients $\mathbb{R}^3$ in its canonical orientation and which, on any positively oriented orthonormal frame, integrates to give Lebesgue measure: $\int_A f\, \Omega = \int_A f\, d\mathcal{L}^3$ for every Borel set $A \subseteq \mathbb{R}^3$ and every integrable $f$. Define the **flux $2$-form** $\omega \in \Omega^2(U)$ associated to $F$ as the contraction \begin{align*} \omega := \iota_F\, \Omega, \end{align*} that is, $\omega_x(v, w) = \Omega_x(F(x), v, w)$ for $x \in U$ and $v, w \in T_x \mathbb{R}^3 = \mathbb{R}^3$. Expanding $\Omega(F, \cdot, \cdot)$ using the formula \begin{align*} \iota_{e_i}(dx_j \wedge dx_k \wedge dx_l) = \delta_{ij}\, dx_k \wedge dx_l - \delta_{ik}\, dx_j \wedge dx_l + \delta_{il}\, dx_j \wedge dx_k \end{align*} and linearity in $F = \sum_i F_i\, \partial_{x_i}$, we obtain the coordinate expression \begin{align*} \omega = F_1\, dx_2 \wedge dx_3 + F_2\, dx_3 \wedge dx_1 + F_3\, dx_1 \wedge dx_2. \end{align*} Since $F_1, F_2, F_3 \in C^\infty(U)$, the form $\omega$ is smooth on $U$. [guided] We need to convert an equality between integrals of a vector field $F$ (one over a $3$-dimensional region, the other over a $2$-dimensional surface) into an equality between integrals of differential forms, so that the powerful machinery of [Stokes' theorem](/theorems/1530) becomes applicable. Differential forms are the right object because their integrals are intrinsically oriented and naturally pull back under boundary inclusions. The recipe is standard: starting from the volume form $\Omega = dx_1 \wedge dx_2 \wedge dx_3$ on $\mathbb{R}^3$, we contract one slot with $F$. The resulting $2$-form $\omega = \iota_F\, \Omega$ is defined by feeding $F$ into the first argument of $\Omega$: \begin{align*} \omega_x(v, w) = \Omega_x(F(x), v, w), \qquad x \in U,\ v, w \in \mathbb{R}^3. \end{align*} Geometrically, $\omega_x(v, w)$ is the signed volume of the parallelepiped spanned by $F(x), v, w$ — a quantity that encodes "how much of $F$ pokes transversely through the parallelogram spanned by $v, w$." To extract the coordinate expression, we use the alternating identity $\iota_{\partial_{x_i}}\Omega = (-1)^{i-1} dx_1 \wedge \cdots \widehat{dx_i} \cdots \wedge dx_n$, which for $\mathbb{R}^3$ specialises to \begin{align*} \iota_{\partial_{x_1}}\Omega &= dx_2 \wedge dx_3, \\ \iota_{\partial_{x_2}}\Omega &= -dx_1 \wedge dx_3 = dx_3 \wedge dx_1, \\ \iota_{\partial_{x_3}}\Omega &= dx_1 \wedge dx_2. \end{align*} Writing $F = F_1\, \partial_{x_1} + F_2\, \partial_{x_2} + F_3\, \partial_{x_3}$ and using linearity of the contraction in its first slot: \begin{align*} \omega = \iota_F\, \Omega = F_1\, dx_2 \wedge dx_3 + F_2\, dx_3 \wedge dx_1 + F_3\, dx_1 \wedge dx_2. \end{align*} The cyclic pattern of subscripts is a useful mnemonic: each $F_i$ multiplies the $2$-form built from the *other two* coordinates in cyclic order. Smoothness of $\omega$ on $U$ is immediate from smoothness of each $F_i$. [/guided] [/step] [step:Compute the exterior derivative $d\omega = (\nabla \cdot F)\, \Omega$] Applying the [Coordinate Formula for the Exterior Derivative](/theorems/3564) to each summand of $\omega$: \begin{align*} d(F_1\, dx_2 \wedge dx_3) &= dF_1 \wedge dx_2 \wedge dx_3 = \partial_{x_1} F_1\, dx_1 \wedge dx_2 \wedge dx_3, \\ d(F_2\, dx_3 \wedge dx_1) &= dF_2 \wedge dx_3 \wedge dx_1 = \partial_{x_2} F_2\, dx_1 \wedge dx_2 \wedge dx_3, \\ d(F_3\, dx_1 \wedge dx_2) &= dF_3 \wedge dx_1 \wedge dx_2 = \partial_{x_3} F_3\, dx_1 \wedge dx_2 \wedge dx_3, \end{align*} where the remaining terms from $dF_i = \sum_j \partial_{x_j} F_i\, dx_j$ vanish because they contain a repeated $dx_j$. Summing, \begin{align*} d\omega = \bigl(\partial_{x_1} F_1 + \partial_{x_2} F_2 + \partial_{x_3} F_3\bigr)\, dx_1 \wedge dx_2 \wedge dx_3 = (\nabla \cdot F)\, \Omega. \end{align*} This is the content of [Exterior Derivative Recovers Gradient, Curl, and Divergence on $\mathbb{R}^3$](/theorems/3566): the [exterior derivative](/theorems/1525) of the flux $2$-form recovers the divergence times the volume form. [guided] We now compute $d\omega$ explicitly. The strategy is to apply the Leibniz rule $d(f\, \alpha) = df \wedge \alpha + f\, d\alpha$ to each summand of $\omega$, then exploit the fact that $d(dx_j \wedge dx_k) = 0$ and that any wedge product containing a repeated $dx_i$ is zero. For the first summand: \begin{align*} d(F_1\, dx_2 \wedge dx_3) = dF_1 \wedge dx_2 \wedge dx_3 = \sum_{j=1}^{3} \partial_{x_j} F_1\, dx_j \wedge dx_2 \wedge dx_3. \end{align*} The terms with $j = 2$ or $j = 3$ contain a repeated $dx_j$ and vanish, leaving only $j = 1$: \begin{align*} d(F_1\, dx_2 \wedge dx_3) = \partial_{x_1} F_1\, dx_1 \wedge dx_2 \wedge dx_3. \end{align*} The same computation for $F_2\, dx_3 \wedge dx_1$ and $F_3\, dx_1 \wedge dx_2$ yields: \begin{align*} d(F_2\, dx_3 \wedge dx_1) &= \partial_{x_2} F_2\, dx_2 \wedge dx_3 \wedge dx_1 = \partial_{x_2} F_2\, dx_1 \wedge dx_2 \wedge dx_3, \\ d(F_3\, dx_1 \wedge dx_2) &= \partial_{x_3} F_3\, dx_3 \wedge dx_1 \wedge dx_2 = \partial_{x_3} F_3\, dx_1 \wedge dx_2 \wedge dx_3, \end{align*} where in each line the second equality is a cyclic re-ordering using $dx_a \wedge dx_b \wedge dx_c = dx_1 \wedge dx_2 \wedge dx_3$ whenever $(a, b, c)$ is an even permutation of $(1, 2, 3)$ — and $(2, 3, 1)$, $(3, 1, 2)$ are both even, so no sign appears. Summing the three: \begin{align*} d\omega = (\partial_{x_1} F_1 + \partial_{x_2} F_2 + \partial_{x_3} F_3)\, \Omega = (\nabla \cdot F)\, \Omega. \end{align*} This is the key identity that makes the Divergence Theorem a special case of [Stokes' Theorem](/theorems/1530): the divergence operator on vector fields corresponds — under the identification $F \leftrightarrow \iota_F\, \Omega$ — to the [exterior derivative](/theorems/1525) on $2$-forms. The same correspondence identifies gradient with $d$ on $0$-forms and curl with $d$ on $1$-forms; the resulting three identities are the content of [Exterior Derivative Recovers Gradient, Curl, and Divergence on $\mathbb{R}^3$](/theorems/3566). [/guided] [/step] [step:Identify the boundary pullback $j^*\omega$ with the flux density $(F \cdot n)\, dS$] Let $j: \partial M \hookrightarrow \mathbb{R}^3$ denote the smooth inclusion of the boundary, and let $dS \in \Omega^2(\partial M)$ denote the Riemannian volume form on $\partial M$ induced by the standard Euclidean metric on $\mathbb{R}^3$ together with the boundary orientation. We claim \begin{align*} j^*\omega = (F \cdot n)|_{\partial M}\, dS \qquad \text{in } \Omega^2(\partial M). \end{align*} Both sides are smooth $2$-forms on the $2$-manifold $\partial M$; it suffices to evaluate each on an arbitrary positively oriented orthonormal frame at an arbitrary point. Fix $p \in \partial M$ and let $(e_1, e_2)$ be a positively oriented orthonormal basis of $T_p \partial M$. By the boundary orientation convention (outward normal first), the triple $(n(p), e_1, e_2)$ is a positively oriented orthonormal basis of $\mathbb{R}^3$. **Right-hand side.** Since $dS$ is the Riemannian volume form induced by the metric and the boundary orientation, $dS_p(e_1, e_2) = 1$. Hence \begin{align*} \bigl((F \cdot n)|_{\partial M}\, dS\bigr)_p (e_1, e_2) = F(p) \cdot n(p). \end{align*} **Left-hand side.** By definition of pullback, \begin{align*} (j^*\omega)_p(e_1, e_2) = \omega_p(e_1, e_2) = \Omega_p(F(p), e_1, e_2) = \det\bigl[F(p)\ \big|\ e_1\ \big|\ e_2\bigr], \end{align*} where the determinant is taken of the $3 \times 3$ matrix with the indicated column vectors (this uses that $\Omega$ evaluated on three vectors is the determinant of the matrix formed by stacking them as columns). Decompose $F(p)$ in the orthonormal basis $(n(p), e_1, e_2)$: \begin{align*} F(p) = (F(p) \cdot n(p))\, n(p) + (F(p) \cdot e_1)\, e_1 + (F(p) \cdot e_2)\, e_2. \end{align*} By multilinearity and antisymmetry of the determinant, only the component along $n(p)$ contributes (the other two columns would repeat $e_1$ or $e_2$): \begin{align*} \det\bigl[F(p)\ \big|\ e_1\ \big|\ e_2\bigr] = (F(p) \cdot n(p))\, \det\bigl[n(p)\ \big|\ e_1\ \big|\ e_2\bigr] = F(p) \cdot n(p), \end{align*} the last equality because $(n(p), e_1, e_2)$ is positively oriented orthonormal, so its determinant equals $+1$. Both sides agree pointwise on every positively oriented orthonormal frame, hence $j^*\omega = (F \cdot n)|_{\partial M}\, dS$ as elements of $\Omega^2(\partial M)$. [guided] The pullback $j^*\omega$ is the restriction of $\omega$ to vectors tangent to $\partial M$. To make the comparison concrete we evaluate both forms on a single positively oriented orthonormal frame at an arbitrary point $p \in \partial M$; since $\partial M$ is $2$-dimensional and both sides are $2$-forms, agreement on one such frame forces agreement as forms. **Setup.** Fix $p \in \partial M$ and pick a positively oriented orthonormal basis $(e_1, e_2)$ of the $2$-dimensional tangent space $T_p \partial M$. The boundary orientation on $\partial M$ is defined by the convention: a frame $(e_1, e_2)$ on $T_p \partial M$ is positively oriented iff $(n(p), e_1, e_2)$ is a positively oriented frame on $T_p \mathbb{R}^3 = \mathbb{R}^3$. So our choice automatically makes $(n(p), e_1, e_2)$ orthonormal and positively oriented in $\mathbb{R}^3$. **Right-hand side.** The Riemannian volume form $dS$ on a $2$-manifold equipped with a metric and orientation is characterised by $dS_p(f_1, f_2) = 1$ on any positively oriented orthonormal basis $(f_1, f_2)$. Thus $dS_p(e_1, e_2) = 1$ and \begin{align*} \bigl((F \cdot n)|_{\partial M}\, dS\bigr)_p (e_1, e_2) = (F(p) \cdot n(p)) \cdot 1 = F(p) \cdot n(p). \end{align*} **Left-hand side.** Pullback acts on forms by precomposition with the differential: $(j^*\omega)_p(v, w) = \omega_p(dj_p\, v, dj_p\, w)$, and since $j$ is an inclusion, $dj_p$ is the inclusion $T_p \partial M \hookrightarrow T_p \mathbb{R}^3$. So $(j^*\omega)_p(e_1, e_2) = \omega_p(e_1, e_2)$. By the definition of $\omega = \iota_F\, \Omega$, \begin{align*} \omega_p(e_1, e_2) = \Omega_p(F(p), e_1, e_2). \end{align*} In the standard basis of $\mathbb{R}^3$, $\Omega$ is the determinant: $\Omega(u, v, w) = \det[u\ |\ v\ |\ w]$ where the bracket denotes the matrix with $u, v, w$ as columns. Hence \begin{align*} \omega_p(e_1, e_2) = \det\bigl[F(p)\ \big|\ e_1\ \big|\ e_2\bigr]. \end{align*} To evaluate this determinant we use that $(n(p), e_1, e_2)$ is an orthonormal basis of $\mathbb{R}^3$, so we may decompose \begin{align*} F(p) = \alpha\, n(p) + \beta\, e_1 + \gamma\, e_2, \end{align*} with $\alpha = F(p) \cdot n(p)$, $\beta = F(p) \cdot e_1$, $\gamma = F(p) \cdot e_2$ (orthonormal decomposition). By multilinearity of the determinant in its first column, \begin{align*} \det[F(p)\ |\ e_1\ |\ e_2] = \alpha \det[n(p)\ |\ e_1\ |\ e_2] + \beta \det[e_1\ |\ e_1\ |\ e_2] + \gamma \det[e_2\ |\ e_1\ |\ e_2]. \end{align*} The middle and right determinants vanish because they have a repeated column. The remaining determinant equals $+1$ because $(n(p), e_1, e_2)$ is positively oriented orthonormal in $\mathbb{R}^3$ (the determinant of an orientation-preserving orthonormal frame is $+1$). Therefore \begin{align*} (j^*\omega)_p(e_1, e_2) = \alpha = F(p) \cdot n(p). \end{align*} **Conclusion.** Both sides evaluate to $F(p) \cdot n(p)$ on every positively oriented orthonormal frame at every $p \in \partial M$. Since a $2$-form on a $2$-manifold is determined by its value on one positively oriented orthonormal frame at each point, the two forms agree: \begin{align*} j^*\omega = (F \cdot n)|_{\partial M}\, dS. \end{align*} This is the bridge that turns [Stokes' Theorem](/theorems/1530) (a statement about integrating differential forms) into the Divergence Theorem (a statement about flux). It is also where the choice of *outward* normal is built in — the opposite (inward) normal would reverse the boundary orientation and introduce a sign. [/guided] [/step] [step:Apply Stokes' Theorem and translate form integrals into measure integrals] The hypotheses of the [Generalised Stokes' Theorem](/theorems/3555) are satisfied: $M$ is a compact, oriented smooth $3$-manifold with boundary; $\partial M$ carries the induced boundary orientation; and $\omega \in \Omega^2(U)$ is a smooth $2$-form whose restriction to $M$ is compactly supported (since $M$ itself is compact). The theorem gives \begin{align*} \int_M d\omega = \int_{\partial M} j^*\omega. \end{align*} For the left-hand side, substitute $d\omega = (\nabla \cdot F)\, \Omega$ from Step 2: \begin{align*} \int_M d\omega = \int_M (\nabla \cdot F)\, \Omega. \end{align*} Since $\Omega = dx_1 \wedge dx_2 \wedge dx_3$ is the standard volume form on $\mathbb{R}^3$, integration of a smooth function against $\Omega$ on a Borel subset of $\mathbb{R}^3$ with its standard orientation equals integration against Lebesgue measure $\mathcal{L}^3$ (this is the defining property of $\mathcal{L}^3$ as the Borel measure represented by $\Omega$; see [Divergence and Volume Form](/theorems/2753) for the manifold formulation). Hence \begin{align*} \int_M (\nabla \cdot F)\, \Omega = \int_M (\nabla \cdot F)(x)\, d\mathcal{L}^3(x). \end{align*} For the right-hand side, substitute $j^*\omega = (F \cdot n)|_{\partial M}\, dS$ from Step 3: \begin{align*} \int_{\partial M} j^*\omega = \int_{\partial M} (F \cdot n)\, dS. \end{align*} The Riemannian volume form $dS$ on the embedded $2$-manifold $\partial M \subset \mathbb{R}^3$ represents the $2$-dimensional [Hausdorff measure](/page/Hausdorff%20Measure) $\mathcal{H}^2$ restricted to $\partial M$ (this is the surface area measure on a smooth submanifold, given equivalently by either the metric volume form or the [Hausdorff measure](/page/Hausdorff%20Measure) construction). Hence \begin{align*} \int_{\partial M} (F \cdot n)\, dS = \int_{\partial M} (F \cdot n)(x)\, d\mathcal{H}^2(x). \end{align*} Combining, \begin{align*} \int_M (\nabla \cdot F)(x)\, d\mathcal{L}^3(x) = \int_M d\omega = \int_{\partial M} j^*\omega = \int_{\partial M} (F \cdot n)(x)\, d\mathcal{H}^2(x), \end{align*} which is the Divergence Theorem. [guided] We now assemble the pieces. The [Generalised Stokes' Theorem](/theorems/3555) states that for a compact, oriented smooth $n$-manifold-with-boundary $N$ and a smooth $(n-1)$-form $\eta$ defined on a neighbourhood of $N$ whose restriction to $N$ is compactly supported, \begin{align*} \int_N d\eta = \int_{\partial N} j^*\eta. \end{align*} We verify the hypotheses in our setting: - **Compact oriented smooth $3$-manifold with boundary:** $M$ is given to be a compact, smoothly embedded $3$-dimensional submanifold with boundary in $\mathbb{R}^3$, equipped with the orientation inherited from the standard orientation of $\mathbb{R}^3$. Its boundary $\partial M$ is a smooth closed $2$-manifold carrying the induced boundary orientation. - **Smooth $2$-form on a neighbourhood of $M$:** the flux form $\omega = \iota_F\, \Omega \in \Omega^2(U)$ from Step 1 is smooth on the open neighbourhood $U \supseteq M$. - **Compact support of the restriction to $M$:** $M$ itself is compact, so $\omega|_M$ is automatically compactly supported (in fact, $M$ is its own support). All hypotheses verified, [Stokes' Theorem](/theorems/1530) applies: \begin{align*} \int_M d\omega = \int_{\partial M} j^*\omega. \end{align*} It remains to translate each side into the classical measure-integral form stated in the theorem. **Left-hand side: from form to Lebesgue measure.** By Step 2, $d\omega = (\nabla \cdot F)\, \Omega$. The integral of a top form on an oriented manifold is defined chart-by-chart via the change-of-variables formula. On $\mathbb{R}^3$ with its standard orientation, the volume form $\Omega = dx_1 \wedge dx_2 \wedge dx_3$ acts on any Borel set $A$ and any integrable function $f$ as \begin{align*} \int_A f\, \Omega = \int_A f(x)\, d\mathcal{L}^3(x), \end{align*} which is the defining property of Lebesgue measure as the unique translation-invariant Borel measure assigning unit volume to the unit cube. The same identification on $M$ (which is a Borel subset of $\mathbb{R}^3$ inheriting the standard orientation) gives \begin{align*} \int_M d\omega = \int_M (\nabla \cdot F)(x)\, d\mathcal{L}^3(x). \end{align*} **Right-hand side: from form to surface measure.** By Step 3, $j^*\omega = (F \cdot n)|_{\partial M}\, dS$, where $dS$ is the Riemannian volume form on $\partial M$. For a smooth $2$-dimensional submanifold $\Sigma \subseteq \mathbb{R}^3$ with the induced metric, the Riemannian volume form $dS$ represents the same measure as the restriction $\mathcal{H}^2|_\Sigma$ of the $2$-dimensional [Hausdorff measure](/page/Hausdorff%20Measure) on $\mathbb{R}^3$. This identification is the standard equivalence between the metric definition of surface area (via parametrisations and Jacobian factors) and the [Hausdorff measure](/page/Hausdorff%20Measure) construction. Applied to $\Sigma = \partial M$, \begin{align*} \int_{\partial M} j^*\omega = \int_{\partial M} (F \cdot n)(x)\, d\mathcal{H}^2(x). \end{align*} **Conclusion.** Chaining the three equalities — translation of the bulk integral, [Stokes' Theorem](/theorems/1530), and translation of the boundary integral — \begin{align*} \int_M (\nabla \cdot F)(x)\, d\mathcal{L}^3(x) &= \int_M d\omega && \text{(Step 2 + form-vs-measure identification)} \\ &= \int_{\partial M} j^*\omega && \text{(Generalised Stokes' Theorem)} \\ &= \int_{\partial M} (F \cdot n)(x)\, d\mathcal{H}^2(x) && \text{(Step 3 + surface form-vs-measure identification)}. \end{align*} This is precisely the Divergence Theorem. [/guided] [/step]