diff --git a/rounding.scad b/rounding.scad
index 0d60c04..b9ef299 100644
--- a/rounding.scad
+++ b/rounding.scad
@@ -76,7 +76,7 @@ include <structs.scad>
 //   ignored.  Note that $fn is interpreted as the number of points on the roundover curve, which is
 //   not equivalent to its meaning for rounding circles because roundovers are usually small fractions
 //   of a circular arc.  When doing continuous curvature rounding be sure to use lots of segments or the effect
-//   will be hidden by the discretization.  Note that if you use $fn then $fn with "smooth" then $fn points are added at each corner, even
+//   will be hidden by the discretization.  Note that if you use $fn with "smooth" then $fn points are added at each corner, even
 //   if the "corner" is flat, with collinear points, so this guarantees a specific output length.  
 //
 // Figure(2D,Med):
diff --git a/transforms.scad b/transforms.scad
index 471198b..0fa9374 100644
--- a/transforms.scad
+++ b/transforms.scad
@@ -349,17 +349,18 @@ function up(z=0, p) = move([0,0,z],p=p);
 //   * `rot(30)` or `rot(a=30)` rotates 30 degrees around the Z axis.
 //   * `rot([20,30,40])` or `rot(a=[20,30,40])` rotates 20 degrees around the X axis, then 30 degrees around the Y axis, then 40 degrees around the Z axis.
 //   * `rot(30, [1,1,0])` or `rot(a=30, v=[1,1,0])` rotates 30 degrees around the axis vector `[1,1,0]`.
-//   * `rot(from=[0,0,1], to=[1,0,0])` rotates the top towards the right, similar to `rot(a=90,v=[0,1,0]`.
-//   * `rot(from=[0,0,1], to=[1,1,0], a=45)` rotates 45 degrees around the Z axis, then rotates the top towards the back-right.  Similar to `rot(a=90,v=[-1,1,0])`
-//   If the `cp` centerpoint argument is given, then rotations are performed around that centerpoint.
-//   If the `reverse` argument is true, then the rotations performed will be exactly reversed.
+//   * `rot(from=[0,0,1], to=[1,0,0])` rotates the `from` vector to line up with the `to` vector, in this case the top to the right and hence equivalent to `rot(a=90,v=[0,1,0]`.
+//   * `rot(from=[0,1,1], to=[1,1,0], a=45)` rotates 45 degrees around the `from` vector ([0,1,1]) and then rotates the `from` vector to align with the `to` vector.  Equivalent to `rot(from=[0,1,1],to=[1,1,0]) rot(a=45,v=[0,1,1])`.  You can also regard `a` as as post-rotation around the `to` vector.  For this form, `a` must be a scalar.  
+//   * If the `cp` centerpoint argument is given, then rotations are performed around that centerpoint.  So `rot(args...,cp=[1,2,3])` is equivalent to `move(-[1,2,3])rot(args...)move([1,2,3])`.
+//   * If the `reverse` argument is true, then the rotations performed will be exactly reversed.
+//   .
 //   The behavior and return value varies depending on how `rot()` is called:
 //   * Called as a module, rotates all children.
 //   * Called as a function with a `p` argument containing a point, returns the rotated point.
 //   * Called as a function with a `p` argument containing a list of points, returns the list of rotated points.
 //   * Called as a function with a [bezier patch](beziers.scad) in the `p` argument, returns the rotated patch.
 //   * Called as a function with a [VNF structure](vnf.scad) in the `p` argument, returns the rotated VNF.
-//   * Called as a function without a `p` argument, and `planar` is true, returns the affine2d rotational matrix.  Requires that `a` is a finite scalar.
+//   * Called as a function without a `p` argument, and `planar` is true, returns the affine2d rotational matrix.  The angle `a` must be a scalar. 
 //   * Called as a function without a `p` argument, and `planar` is false, returns the affine3d rotational matrix.
 //
 // Arguments: