[proofplan] We prove the operator identity $\Delta\alpha = \nabla^*\nabla\alpha + \operatorname{Ric}(\alpha)$ on $\Omega^1(M)$ pointwise: at each $p \in M$ we choose a normal orthonormal frame $\{e_k\}$ — meaning $\nabla_Y e_k|_p = 0$ for every $Y$ and every $k$ — which makes covariant derivatives at $p$ behave like ordinary derivatives. We expand $\Delta = d\delta + \delta d$ on a $1$-form $\alpha$ in that frame, separating $\langle d\delta\alpha, X\rangle$ and $\langle \delta d\alpha, X\rangle$. Adding the two halves, the second covariant derivatives reorganise into $\langle \nabla^*\nabla\alpha, X\rangle$ plus a commutator $[\nabla_{e_k}, \nabla_X] - \nabla_{[e_k,X]}$ acting on $\alpha$. The Ricci identity for $1$-forms identifies that commutator with the Riemann curvature, and tracing over the frame collapses the curvature term to the Ricci endomorphism $\operatorname{Ric}(\alpha)$. The chapter sign convention $R(X,Y) = -[\nabla_X,\nabla_Y] + \nabla_{[X,Y]}$ on vector fields (equivalently $R = -\nabla \circ \nabla$ on the algebra of forms) is used throughout to fix signs. [/proofplan]

[step:Set up the pointwise computation in a normal orthonormal frame] Let $(M, g)$ be an oriented Riemannian manifold of dimension $n$ with Levi-Civita connection $\nabla$, and let $\alpha \in \Omega^1(M)$. Fix $p \in M$ and let $X \in \mathfrak{X}(M)$. We prove the identity \begin{align*} \langle \Delta\alpha, X\rangle\big|_p = \langle \nabla^*\nabla\alpha, X\rangle\big|_p + \operatorname{Ric}(\alpha)(X)\big|_p. \end{align*} Since $p$ and $X$ are arbitrary, the operator identity $\Delta\alpha = \nabla^*\nabla\alpha + \operatorname{Ric}(\alpha)$ on $\Omega^1(M)$ follows. Choose a **normal orthonormal frame field** $\{e_1, \ldots, e_n\}$ at $p$: a smooth orthonormal frame on a neighbourhood $U \subseteq M$ of $p$ satisfying $\nabla_Y e_k|_p = 0$ for all $Y \in T_p M$ and all $k = 1, \ldots, n$. Such a frame is constructed by choosing an orthonormal basis of $T_p M$ and parallel-transporting it along radial geodesics in normal coordinates centred at $p$; the resulting frame is smooth on a normal neighbourhood and its covariant derivatives at $p$ vanish (this is the defining property of normal frames at a point). In such a frame at $p$: \begin{align*} [e_k, X]\big|_p &= \nabla_{e_k} X\big|_p - \nabla_X e_k\big|_p = \nabla_{e_k} X\big|_p, \end{align*} since $\nabla_X e_k|_p = 0$ by the normal frame condition. All identities below are derived at $p$; tensoriality of both sides shows they extend to all of $M$. We use the chapter sign convention $\Delta = d\delta + \delta d$ for the Hodge Laplacian on $1$-forms, and $R(X, Y) = -[\nabla_X, \nabla_Y] + \nabla_{[X, Y]}$ for the curvature operator on vector fields, equivalent to $R = -\nabla \circ \nabla$ on the form algebra. The dual statement on $1$-forms is the [Curvature as Commutator of Covariant Derivatives](/theorems/2703) identity: \begin{align*} \big([\nabla_X, \nabla_Y] - \nabla_{[X, Y]}\big)\alpha = -R(X, Y)\alpha, \end{align*} where $R(X, Y)$ acts on $\alpha \in \Omega^1(M)$ as the dual of its action on vector fields: $\big(R(X,Y)\alpha\big)(Z) = -\alpha(R(X,Y)Z)$. We unpack this dual action when needed. [/step]

[step:Expand $\langle d\delta\alpha, X\rangle$ in the frame] By the [Local Formula for $\delta$ on 1-Forms](/theorems/2756), \begin{align*} \delta\alpha = -\sum_{i=1}^n (\nabla_{e_i}\alpha)(e_i) = -\sum_{i=1}^n \langle \nabla_{e_i}\alpha, e_i\rangle, \end{align*} where $\langle\cdot,\cdot\rangle$ denotes the metric pairing of a $1$-form with a vector field. Since $\delta\alpha$ is a smooth function, $d(\delta\alpha)$ is a $1$-form whose pairing with $X$ is \begin{align*} \langle d\delta\alpha, X\rangle = X(\delta\alpha) = -\sum_{i=1}^n X\big(\langle \nabla_{e_i}\alpha, e_i\rangle\big). \end{align*} Apply metric compatibility of $\nabla$ to differentiate the pairing: \begin{align*} X\big(\langle \nabla_{e_i}\alpha, e_i\rangle\big) = \langle \nabla_X \nabla_{e_i}\alpha, e_i\rangle + \langle \nabla_{e_i}\alpha, \nabla_X e_i\rangle. \end{align*} At $p$ the second term vanishes because $\nabla_X e_i|_p = 0$. Hence at $p$: \begin{align*} \langle d\delta\alpha, X\rangle\big|_p = -\sum_{i=1}^n \langle \nabla_X \nabla_{e_i}\alpha, e_i\rangle\big|_p. \tag{A} \end{align*} [/step]

[step:Expand $\langle \delta d\alpha, X\rangle$ in the frame] Since the Levi-Civita connection is torsion-free, $[X, Y] = \nabla_X Y - \nabla_Y X$, and the standard formula for the exterior derivative of a $1$-form gives \begin{align*} d\alpha(X, Y) = X(\alpha(Y)) - Y(\alpha(X)) - \alpha([X, Y]) = \langle \nabla_X\alpha, Y\rangle - \langle \nabla_Y\alpha, X\rangle, \end{align*} the second equality following from metric compatibility (the cross-terms involving $\nabla_X Y$ and $\nabla_Y X$ cancel against $\alpha([X,Y]) = \alpha(\nabla_X Y) - \alpha(\nabla_Y X)$). By the [Local Formula for $\delta$ on 2-Forms](/theorems/2757), for any $\beta \in \Omega^2(M)$ and $X \in \mathfrak{X}(M)$, \begin{align*} (\delta\beta)(X) = -\sum_{k=1}^n (\nabla_{e_k}\beta)(e_k, X). \end{align*} Apply this to $\beta = d\alpha$: \begin{align*} (\delta d\alpha)(X) = -\sum_{k=1}^n (\nabla_{e_k} d\alpha)(e_k, X). \end{align*} Since $d\alpha$ is a $2$-tensor, the covariant derivative satisfies the Leibniz rule: \begin{align*} (\nabla_{e_k} d\alpha)(e_k, X) = e_k\big(d\alpha(e_k, X)\big) - d\alpha(\nabla_{e_k} e_k, X) - d\alpha(e_k, \nabla_{e_k} X). \end{align*} At $p$ the term $\nabla_{e_k} e_k|_p$ vanishes, so \begin{align*} (\nabla_{e_k} d\alpha)(e_k, X)\big|_p = e_k\big(d\alpha(e_k, X)\big)\big|_p - d\alpha(e_k, \nabla_{e_k} X)\big|_p. \end{align*} Substitute the formula $d\alpha(Y, Z) = \langle \nabla_Y\alpha, Z\rangle - \langle \nabla_Z\alpha, Y\rangle$ and apply metric compatibility to expand each piece. We compute $e_k(d\alpha(e_k, X))$ and $d\alpha(e_k, \nabla_{e_k}X)$ separately, then subtract. For the first piece, $d\alpha(e_k, X) = \langle \nabla_{e_k}\alpha, X\rangle - \langle \nabla_X\alpha, e_k\rangle$. Differentiating each pairing along $e_k$ by metric compatibility produces four terms: \begin{align*} e_k\big(\langle \nabla_{e_k}\alpha, X\rangle\big) &= \langle \nabla_{e_k}\nabla_{e_k}\alpha, X\rangle + \langle \nabla_{e_k}\alpha, \nabla_{e_k} X\rangle, \\ e_k\big(\langle \nabla_X\alpha, e_k\rangle\big) &= \langle \nabla_{e_k}\nabla_X\alpha, e_k\rangle + \langle \nabla_X\alpha, \nabla_{e_k} e_k\rangle. \end{align*} At $p$ the term $\langle \nabla_X\alpha, \nabla_{e_k} e_k\rangle$ vanishes because $\nabla_{e_k} e_k|_p = 0$ for the normal frame. Subtracting, \begin{align*} e_k\big(d\alpha(e_k, X)\big)\big|_p = \langle \nabla_{e_k}\nabla_{e_k}\alpha, X\rangle\big|_p + \langle \nabla_{e_k}\alpha, \nabla_{e_k} X\rangle\big|_p - \langle \nabla_{e_k}\nabla_X\alpha, e_k\rangle\big|_p. \tag{$*$} \end{align*} For the second piece, applying the formula for $d\alpha$ directly: \begin{align*} d\alpha(e_k, \nabla_{e_k} X) = \langle \nabla_{e_k}\alpha, \nabla_{e_k} X\rangle - \langle \nabla_{\nabla_{e_k} X}\alpha, e_k\rangle. \tag{$**$} \end{align*} Now $(\nabla_{e_k} d\alpha)(e_k, X)|_p = e_k(d\alpha(e_k, X))|_p - d\alpha(e_k, \nabla_{e_k} X)|_p$. The cross-term $\langle \nabla_{e_k}\alpha, \nabla_{e_k} X\rangle$ appears in $(*)$ with a plus sign and in $(**)$ with a plus sign, so it cancels exactly upon subtraction. We are left with \begin{align*} (\nabla_{e_k} d\alpha)(e_k, X)\big|_p = \langle \nabla_{e_k}\nabla_{e_k}\alpha, X\rangle\big|_p - \langle \nabla_{e_k}\nabla_X\alpha, e_k\rangle\big|_p + \langle \nabla_{\nabla_{e_k} X}\alpha, e_k\rangle\big|_p. \end{align*} Thus, in summary: of the four terms generated by metric compatibility, the $\nabla_{e_k}e_k$-term vanishes at $p$ because the frame is normal, and the cross-term $\langle \nabla_{e_k}\alpha, \nabla_{e_k} X\rangle$ cancels against the matching contribution from $d\alpha(e_k, \nabla_{e_k} X)$. Summing over $k$ and inserting the overall minus sign from the formula for $\delta$: \begin{align*} \langle \delta d\alpha, X\rangle\big|_p = -\sum_{k=1}^n \langle \nabla_{e_k}\nabla_{e_k}\alpha, X\rangle\big|_p + \sum_{k=1}^n \langle \nabla_{e_k}\nabla_X\alpha, e_k\rangle\big|_p - \sum_{k=1}^n \langle \nabla_{\nabla_{e_k}X}\alpha, e_k\rangle\big|_p. \tag{B} \end{align*} The first sum on the right of (B) is exactly $\langle \nabla^*\nabla\alpha, X\rangle$ at $p$: by the [Adjoint Formula for Covariant Derivative](/theorems/2758), $\nabla^*\nabla\alpha = -\sum_i (\nabla_{e_i}\nabla_{e_i}\alpha - \nabla_{\nabla_{e_i}e_i}\alpha)$, and at $p$ the frame term $\nabla_{e_i}e_i|_p = 0$, so $\nabla^*\nabla\alpha|_p = -\sum_i \nabla_{e_i}\nabla_{e_i}\alpha|_p$. [/step]

[step:Add the two halves and isolate the curvature commutator] Adding (A) and (B): \begin{align*} \langle \Delta\alpha, X\rangle\big|_p = \langle \nabla^*\nabla\alpha, X\rangle\big|_p + \sum_{k=1}^n \big\langle \big([\nabla_{e_k}, \nabla_X] - \nabla_{\nabla_{e_k}X}\big)\alpha, e_k\big\rangle\big|_p, \end{align*} where we have grouped the second-derivative terms: $-\langle \nabla_X\nabla_{e_k}\alpha, e_k\rangle + \langle \nabla_{e_k}\nabla_X\alpha, e_k\rangle = \langle [\nabla_{e_k}, \nabla_X]\alpha, e_k\rangle$, and the third sum in (B) contributes $-\langle \nabla_{\nabla_{e_k}X}\alpha, e_k\rangle$. At $p$, the normal frame condition gives $[e_k, X]|_p = \nabla_{e_k}X|_p - \nabla_X e_k|_p = \nabla_{e_k}X|_p$, so $\nabla_{\nabla_{e_k}X}\alpha|_p = \nabla_{[e_k, X]}\alpha|_p$. Substituting: \begin{align*} \langle \Delta\alpha, X\rangle\big|_p = \langle \nabla^*\nabla\alpha, X\rangle\big|_p + \sum_{k=1}^n \big\langle \big([\nabla_{e_k}, \nabla_X] - \nabla_{[e_k, X]}\big)\alpha, e_k\big\rangle\big|_p. \tag{C} \end{align*} [/step]

[step:Apply the Ricci identity for $1$-forms to convert the commutator to curvature] We derive the dual Ricci identity on $1$-forms from the vector-field statement and metric compatibility. The curvature $R$ extends from $\mathfrak{X}(M)$ to the full tensor algebra as a derivation commuting with metric contractions; this extension is what gives $R(X, Y)$ a meaning when acting on a $1$-form. Concretely, by metric compatibility applied to the duality pairing $\alpha(Z) = g(X_\alpha, Z)$, we have for any $X, Y \in \mathfrak{X}(M)$, any $1$-form $\alpha$, and any vector field $Z$, \begin{align*} X\big(\alpha(Z)\big) = (\nabla_X \alpha)(Z) + \alpha(\nabla_X Z). \end{align*} Differentiating again along $Y$, doing the same with $X$ and $Y$ swapped, and antisymmetrising, the second-derivative terms applied to the scalar $\alpha(Z)$ reorganise into commutators acting on $\alpha$ and on $Z$ separately, while the Lie-bracket adjustment supplies the $\nabla_{[X, Y]}$ terms. Matching coefficients yields the derivation identity \begin{align*} \big([\nabla_X, \nabla_Y] - \nabla_{[X, Y]}\big)\alpha\,(Z) + \alpha\big(([\nabla_X, \nabla_Y] - \nabla_{[X, Y]})Z\big) = 0. \end{align*} Under the chapter sign convention $R(X, Y) = -[\nabla_X, \nabla_Y] + \nabla_{[X, Y]}$ on vector fields (equivalently $R = -\nabla \circ \nabla$ on the form algebra), the inner commutator acting on $Z$ equals $-R(X, Y)Z$. Substituting and rearranging, \begin{align*} \big([\nabla_X, \nabla_Y] - \nabla_{[X, Y]}\big)\alpha\,(Z) = \alpha\big(R(X, Y)Z\big). \end{align*} Defining the dual action of $R(X, Y)$ on $1$-forms by the derivation rule on the duality pairing, \begin{align*} \big(R(X, Y)\alpha\big)(Z) := -\alpha\big(R(X, Y)Z\big), \end{align*} we obtain the dual statement of the [Curvature as Commutator of Covariant Derivatives](/theorems/2703) on $1$-forms: \begin{align*} \big([\nabla_X, \nabla_Y] - \nabla_{[X, Y]}\big)\alpha = -R(X, Y)\alpha. \end{align*} The minus sign on the right is the dual of the sign in $R = -\nabla \circ \nabla$ on vector fields: extending $R$ as a derivation on the tensor algebra forces the sign to flip when passing through the duality pairing. Pairing with $e_k$: \begin{align*} \big\langle ([\nabla_{e_k}, \nabla_X] - \nabla_{[e_k, X]})\alpha, e_k\big\rangle = \big(-R(e_k, X)\alpha\big)(e_k) = -\big(-\alpha(R(e_k, X)e_k)\big) = \alpha\big(R(e_k, X) e_k\big). \end{align*} Summing over $k$: \begin{align*} \sum_{k=1}^n \big\langle ([\nabla_{e_k}, \nabla_X] - \nabla_{[e_k, X]})\alpha, e_k\big\rangle\big|_p = \sum_{k=1}^n \alpha\big(R(e_k, X) e_k\big)\big|_p. \tag{D} \end{align*} Now identify the right-hand side with the Ricci endomorphism. Let $X_\alpha \in \mathfrak{X}(M)$ be the metric dual of $\alpha$ — the unique vector field with $\alpha(Z) = g(X_\alpha, Z)$ for all $Z$. We rewrite $\alpha(R(e_k, X)e_k)$ using the [Symmetries of the Riemann Curvature Tensor](/theorems/2704), which provide both the pair-symmetry $R_{ab,cd} = R_{cd,ab}$ and the antisymmetry-in-the-last-two-slots $R_{ab,cd} = -R_{ab,dc}$. Writing $R(W, Y, Z, V) := g(R(W, Y) Z, V)$ for the fully covariant $(0, 4)$-tensor, the chain is: \begin{align*} \alpha\big(R(e_k, X)e_k\big) &= g\big(X_\alpha, R(e_k, X)e_k\big) && \text{(metric duality)} \\ &= g\big(R(e_k, X)e_k, X_\alpha\big) && \text{(symmetry of $g$)} \\ &= R(e_k, X, e_k, X_\alpha) && \text{(notation for $R_{ab,cd}$)} \\ &= R(e_k, X_\alpha, e_k, X) && \text{(pair-symmetry $R_{ab,cd} = R_{cd,ab}$)} \\ &= g\big(R(e_k, X_\alpha)e_k, X\big) && \text{(notation back).} \end{align*} Both symmetries are needed: pair-symmetry $R_{ab,cd} = R_{cd,ab}$ swaps the first pair $(a,b) = (e_k, X)$ with the second pair $(c,d) = (e_k, X_\alpha)$, and antisymmetry-in-the-last-two-slots $R_{ab,cd} = -R_{ab,dc}$ ensures $(c,d)$ transforms as a coherent block under that swap. Pair-symmetry on a Levi-Civita connection follows from the algebraic Bianchi identity together with the antisymmetries in the first and last pairs of indices. Summing, \begin{align*} \sum_{k=1}^n \alpha\big(R(e_k, X) e_k\big)\big|_p = \sum_{k=1}^n g\big(R(e_k, X_\alpha) e_k, X\big)\big|_p = \operatorname{Ric}(X_\alpha, X)\big|_p, \end{align*} the last equality by the definition of Ricci curvature as the trace $\operatorname{Ric}(Y, Z) = \sum_{k=1}^n g(R(e_k, Y) e_k, Z)$ over an orthonormal frame, with the chapter sign convention. By definition of the Ricci endomorphism on $1$-forms, $\operatorname{Ric}(\alpha)(X) = \operatorname{Ric}(X_\alpha, X)$, so \begin{align*} \sum_{k=1}^n \alpha\big(R(e_k, X) e_k\big)\big|_p = \operatorname{Ric}(\alpha)(X)\big|_p. \tag{E} \end{align*} [/step]

[step:Combine to obtain the Bochner–Weitzenböck identity]Substituting (E) and (D) into (C): \begin{align*} \langle \Delta\alpha, X\rangle\big|_p = \langle \nabla^*\nabla\alpha, X\rangle\big|_p + \operatorname{Ric}(\alpha)(X)\big|_p. \end{align*} This is the equivalent vector-field-paired form of the theorem at $p$. Both sides are tensorial in $\alpha$ and $X$, and $p$ is arbitrary, so the pointwise identity extends to a global operator identity: \begin{align*} \Delta\alpha = \nabla^*\nabla\alpha + \operatorname{Ric}(\alpha) \qquad \text{on } \Omega^1(M). \end{align*}[/step]

[guided]We have done all the hard work; only the bookkeeping remains. The previous steps produced three labelled identities at $p$: equation (C), which expresses $\langle \Delta\alpha, X\rangle|_p$ as $\langle \nabla^*\nabla\alpha, X\rangle|_p$ plus a sum of commutator-type terms; equation (D), which converts that commutator sum into $\sum_k \alpha(R(e_k, X)e_k)|_p$ using the dual Ricci identity on $1$-forms; and equation (E), which traces that curvature sum over the frame and identifies it with the Ricci endomorphism applied to $\alpha$. We now plug (E) into (D) into (C) and check that the result extends from $p$ to a global tensor identity. **Substituting (E) and (D) into (C).** Recall: \begin{align*} \langle \Delta\alpha, X\rangle\big|_p &= \langle \nabla^*\nabla\alpha, X\rangle\big|_p + \sum_{k=1}^n \big\langle \big([\nabla_{e_k}, \nabla_X] - \nabla_{[e_k, X]}\big)\alpha, e_k\big\rangle\big|_p && \text{(C)}, \\ \sum_{k=1}^n \big\langle \big([\nabla_{e_k}, \nabla_X] - \nabla_{[e_k, X]}\big)\alpha, e_k\big\rangle\big|_p &= \sum_{k=1}^n \alpha\big(R(e_k, X) e_k\big)\big|_p && \text{(D)}, \\ \sum_{k=1}^n \alpha\big(R(e_k, X) e_k\big)\big|_p &= \operatorname{Ric}(\alpha)(X)\big|_p && \text{(E)}. \end{align*} Chaining (C) through (D) through (E): \begin{align*} \langle \Delta\alpha, X\rangle\big|_p = \langle \nabla^*\nabla\alpha, X\rangle\big|_p + \operatorname{Ric}(\alpha)(X)\big|_p. \end{align*} This is the Bochner–Weitzenböck identity at the single point $p$, paired with an arbitrary tangent vector $X|_p$. **From pointwise to global.** Why does an identity that holds at one point — and was derived using a frame defined only near that point — give an operator identity on all of $M$? The reason is tensoriality. The two sides of \begin{align*} \Delta\alpha = \nabla^*\nabla\alpha + \operatorname{Ric}(\alpha) \end{align*} are both $1$-forms on $M$, built tensorially from $\alpha$ and the metric: $\Delta = d\delta + \delta d$ is defined globally without reference to any frame, $\nabla^*\nabla\alpha$ is the connection Laplacian (also frame-independent), and $\operatorname{Ric}(\alpha)$ is contraction of the curvature tensor — a global tensor — against $\alpha$. The normal orthonormal frame was a computational convenience used to produce a valid value at $p$; the fact that the $\nabla_Y e_k|_p = 0$ and $\nabla_{e_k} e_k|_p = 0$ terms vanished was scaffolding that disappears once both sides are recognised as values of globally defined tensors. Since the identity holds for every $p \in M$ and every $X|_p \in T_p M$, and both sides are smooth tensor fields, the identity holds as an equality of $1$-forms: \begin{align*} \Delta\alpha = \nabla^*\nabla\alpha + \operatorname{Ric}(\alpha) \qquad \text{on } \Omega^1(M). \end{align*} **A sanity check on the sign.** It is worth pausing to confirm that the plus sign in front of $\operatorname{Ric}(\alpha)$ is what we expected from the chapter conventions. In step 4 we used $R = -\nabla \circ \nabla$ on the form algebra to get $([\nabla_X, \nabla_Y] - \nabla_{[X,Y]})\alpha = -R(X, Y)\alpha$, and the dual action $(R(X,Y)\alpha)(Z) = -\alpha(R(X, Y)Z)$ contributed a second minus sign. The two minus signs composed to give $+\alpha(R(e_k, X)e_k)$ in (D). In step 4 we then used pair-symmetry of $R$ to rewrite this as $g(R(e_k, X_\alpha)e_k, X)$ and identified the trace with $\operatorname{Ric}(X_\alpha, X) = \operatorname{Ric}(\alpha)(X)$. No sign was lost or gained in the trace, so the curvature term enters with a plus sign — agreeing with what a coordinate-free derivation under the chapter's conventions predicts. **The arbitrary-$X$ argument.** Because $X$ was arbitrary in $\mathfrak{X}(M)$, the identity $\langle \Delta\alpha, X\rangle = \langle \nabla^*\nabla\alpha + \operatorname{Ric}(\alpha), X\rangle$ holds when paired with any vector field, hence the two $1$-forms $\Delta\alpha$ and $\nabla^*\nabla\alpha + \operatorname{Ric}(\alpha)$ agree at every point: a $1$-form is determined by its evaluation on all vector fields. Combined with arbitrariness of $p$, this completes the upgrade from a pointwise scalar identity to a global identity of $1$-forms. This proves the Bochner–Weitzenböck identity for $1$-forms, exhibiting the Hodge Laplacian as the connection Laplacian plus the Ricci endomorphism — the prototype Weitzenböck formula, and the basis of the Bochner technique for proving vanishing theorems under positive-curvature hypotheses.[/guided]