diff --git a/lapack/gonum/dgesvd.go b/lapack/gonum/dgesvd.go
index 7b0ca66e..6dcfda9c 100644
--- a/lapack/gonum/dgesvd.go
+++ b/lapack/gonum/dgesvd.go
@@ -26,7 +26,7 @@ const noSVDO = "dgesvd: not coded for overwrite"
 // jobU and jobVT are options for computing the singular vectors. The behavior
 // is as follows
 //  jobU == lapack.SVDAll       All m columns of U are returned in u
-//  jobU == lapack.SVDInPlace   The first min(m,n) columns are returned in u
+//  jobU == lapack.SVDStore     The first min(m,n) columns are returned in u
 //  jobU == lapack.SVDOverwrite The first min(m,n) columns of U are written into a
 //  jobU == lapack.SVDNone      The columns of U are not computed.
 // The behavior is the same for jobVT and the rows of V^T. At most one of jobU
@@ -40,12 +40,12 @@ const noSVDO = "dgesvd: not coded for overwrite"
 // values in decreasing order.
 //
 // u contains the left singular vectors on exit, stored column-wise. If
-// jobU == lapack.SVDAll, u is of size m×m. If jobU == lapack.SVDInPlace u is
+// jobU == lapack.SVDAll, u is of size m×m. If jobU == lapack.SVDStore u is
 // of size m×min(m,n). If jobU == lapack.SVDOverwrite or lapack.SVDNone, u is
 // not used.
 //
 // vt contains the left singular vectors on exit, stored row-wise. If
-// jobV == lapack.SVDAll, vt is of size n×m. If jobVT == lapack.SVDInPlace vt is
+// jobV == lapack.SVDAll, vt is of size n×m. If jobVT == lapack.SVDStore vt is
 // of size min(m,n)×n. If jobVT == lapack.SVDOverwrite or lapack.SVDNone, vt is
 // not used.
 //
@@ -61,12 +61,12 @@ func (impl Implementation) Dgesvd(jobU, jobVT lapack.SVDJob, m, n int, a []float
 	checkMatrix(m, n, a, lda)
 	if jobU == lapack.SVDAll {
 		checkMatrix(m, m, u, ldu)
-	} else if jobU == lapack.SVDInPlace {
+	} else if jobU == lapack.SVDStore {
 		checkMatrix(m, minmn, u, ldu)
 	}
 	if jobVT == lapack.SVDAll {
 		checkMatrix(n, n, vt, ldvt)
-	} else if jobVT == lapack.SVDInPlace {
+	} else if jobVT == lapack.SVDStore {
 		checkMatrix(minmn, n, vt, ldvt)
 	}
 	if jobU == lapack.SVDOverwrite && jobVT == lapack.SVDOverwrite {
@@ -83,13 +83,13 @@ func (impl Implementation) Dgesvd(jobU, jobVT lapack.SVDJob, m, n int, a []float
 	}
 
 	wantua := jobU == lapack.SVDAll
-	wantus := jobU == lapack.SVDInPlace
+	wantus := jobU == lapack.SVDStore
 	wantuas := wantua || wantus
 	wantuo := jobU == lapack.SVDOverwrite
 	wantun := jobU == lapack.SVDNone
 
 	wantva := jobVT == lapack.SVDAll
-	wantvs := jobVT == lapack.SVDInPlace
+	wantvs := jobVT == lapack.SVDStore
 	wantvas := wantva || wantvs
 	wantvo := jobVT == lapack.SVDOverwrite
 	wantvn := jobVT == lapack.SVDNone
diff --git a/lapack/lapack.go b/lapack/lapack.go
index 2c94431b..031e3193 100644
--- a/lapack/lapack.go
+++ b/lapack/lapack.go
@@ -108,10 +108,10 @@ const (
 type SVDJob byte
 
 const (
-	SVDAll       SVDJob = 'A' // Compute all singular vectors
-	SVDInPlace   SVDJob = 'S' // Compute the first singular vectors and store them in provided storage.
-	SVDOverwrite SVDJob = 'O' // Compute the singular vectors and store them in input matrix
-	SVDNone      SVDJob = 'N' // Do not compute singular vectors
+	SVDAll       SVDJob = 'A' // Compute all columns of the matrix U or V.
+	SVDStore     SVDJob = 'S' // Compute the singular vectors and store them in the matrix U or V.
+	SVDOverwrite SVDJob = 'O' // Compute the singular vectors and overwrite them on the input matrix A.
+	SVDNone      SVDJob = 'N' // Do not compute singular vectors.
 )
 
 // GSVDJob specifies the singular vector computation type for Generalized SVD.
diff --git a/lapack/lapack64/lapack64.go b/lapack/lapack64/lapack64.go
index 3ca7c15a..03139d21 100644
--- a/lapack/lapack64/lapack64.go
+++ b/lapack/lapack64/lapack64.go
@@ -152,7 +152,7 @@ func Gelqf(a blas64.General, tau, work []float64, lwork int) {
 // jobU and jobVT are options for computing the singular vectors. The behavior
 // is as follows
 //  jobU == lapack.SVDAll       All m columns of U are returned in u
-//  jobU == lapack.SVDInPlace   The first min(m,n) columns are returned in u
+//  jobU == lapack.SVDStore     The first min(m,n) columns are returned in u
 //  jobU == lapack.SVDOverwrite The first min(m,n) columns of U are written into a
 //  jobU == lapack.SVDNone      The columns of U are not computed.
 // The behavior is the same for jobVT and the rows of V^T. At most one of jobU
@@ -166,12 +166,12 @@ func Gelqf(a blas64.General, tau, work []float64, lwork int) {
 // values in decreasing order.
 //
 // u contains the left singular vectors on exit, stored columnwise. If
-// jobU == lapack.SVDAll, u is of size m×m. If jobU == lapack.SVDInPlace u is
+// jobU == lapack.SVDAll, u is of size m×m. If jobU == lapack.SVDStore u is
 // of size m×min(m,n). If jobU == lapack.SVDOverwrite or lapack.SVDNone, u is
 // not used.
 //
 // vt contains the left singular vectors on exit, stored rowwise. If
-// jobV == lapack.SVDAll, vt is of size n×m. If jobVT == lapack.SVDInPlace vt is
+// jobV == lapack.SVDAll, vt is of size n×m. If jobVT == lapack.SVDStore vt is
 // of size min(m,n)×n. If jobVT == lapack.SVDOverwrite or lapack.SVDNone, vt is
 // not used.
 //
diff --git a/lapack/testlapack/dgesvd.go b/lapack/testlapack/dgesvd.go
index d73c2de9..7c991846 100644
--- a/lapack/testlapack/dgesvd.go
+++ b/lapack/testlapack/dgesvd.go
@@ -111,16 +111,16 @@ func DgesvdTest(t *testing.T, impl Dgesvder) {
 		svdCheck(t, false, errStr, m, n, s, a, u, ldu, vt, ldvt, aCopy, lda)
 		svdCheckPartial(t, impl, lapack.SVDAll, errStr, uAllOrig, vtAllOrig, aCopy, m, n, a, lda, s, u, ldu, vt, ldvt, work, false)
 
-		// Test InPlace
-		jobU = lapack.SVDInPlace
-		jobVT = lapack.SVDInPlace
+		// Test SVDStore
+		jobU = lapack.SVDStore
+		jobVT = lapack.SVDStore
 		copy(a, aCopy)
 		copy(u, uAllOrig)
 		copy(vt, vtAllOrig)
 
 		impl.Dgesvd(jobU, jobVT, m, n, a, lda, s, u, ldu, vt, ldvt, work, len(work))
 		svdCheck(t, true, errStr, m, n, s, a, u, ldu, vt, ldvt, aCopy, lda)
-		svdCheckPartial(t, impl, lapack.SVDInPlace, errStr, uAllOrig, vtAllOrig, aCopy, m, n, a, lda, s, u, ldu, vt, ldvt, work, false)
+		svdCheckPartial(t, impl, lapack.SVDStore, errStr, uAllOrig, vtAllOrig, aCopy, m, n, a, lda, s, u, ldu, vt, ldvt, work, false)
 	}
 }
 
diff --git a/mat/svd.go b/mat/svd.go
index ef1f21cf..4600f349 100644
--- a/mat/svd.go
+++ b/mat/svd.go
@@ -26,15 +26,14 @@ type SVD struct {
 //
 // The full singular value decomposition (kind == SVDFull) deconstructs A as
 //  A = U * Σ * V^T
-// where Σ is an m×n diagonal matrix of singular vectors, U is an m×m unitary
+// where Σ is an m×n diagonal matrix of singular values, U is an m×m unitary
 // matrix of left singular vectors, and V is an n×n matrix of right singular vectors.
 //
 // It is frequently not necessary to compute the full SVD. Computation time and
 // storage costs can be reduced using the appropriate kind. Only the singular
 // values can be computed (kind == SVDNone), or a "thin" representation of the
 // singular vectors (kind = SVDThin). The thin representation can save a significant
-// amount of memory if m >> n. See the documentation for the lapack.SVDKind values
-// for more information.
+// amount of memory if m >> n.
 //
 // Factorize returns whether the decomposition succeeded. If the decomposition
 // failed, routines that require a successful factorization will panic.
@@ -81,8 +80,8 @@ func (svd *SVD) Factorize(a Matrix, kind SVDKind) (ok bool) {
 			Stride: n,
 			Data:   use(svd.vt.Data, min(m, n)*n),
 		}
-		jobU = lapack.SVDInPlace
-		jobVT = lapack.SVDInPlace
+		jobU = lapack.SVDStore
+		jobVT = lapack.SVDStore
 	}
 
 	// A is destroyed on call, so copy the matrix.
@@ -119,7 +118,7 @@ func (svd *SVD) Cond() float64 {
 // Values returns the singular values of the factorized matrix in decreasing order.
 // If the input slice is non-nil, the values will be stored in-place into the slice.
 // In this case, the slice must have length min(m,n), and Values will panic with
-// matrix.ErrSliceLengthMismatch otherwise. If the input slice is nil,
+// ErrSliceLengthMismatch otherwise. If the input slice is nil,
 // a new slice of the appropriate length will be allocated and returned.
 //
 // Values will panic if the receiver does not contain a successful factorization.