I’m stuck on reflections when the mirror line isn’t horizontal or vertical. I totally get that a reflection should keep points the same perpendicular distance from the line, but when the line is tilted (like y = 2x + 3), my brain scrambles the direction and I second-guess everything.
Personal backstory: I messed this up on a quiz last year-I tried to reflect a triangle across a line like y = x + 1, and I somehow ended up placing one vertex on the wrong side. Ever since then, I freeze up unless the line is x = c or y = c.
My question: What’s a reliable, quick method to reflect a point (a, b) across a general line y = mx + b without memorizing a giant formula? If a formula is the way to go, can someone explain the idea behind it so it sticks in my head?
Follow-up: If I do two reflections across two different lines, how can I predict whether the overall effect is a rotation or a translation, and how do I figure out the rotation angle or translation vector just from the lines?
















3 Responses
Quick method: shift down by b so the line is y=mx, rotate by θ = arctan(m) to flatten it to y=0, reflect by y → −y, then rotate back and shift up by b. For two reflections, intersecting lines give a rotation by twice the angle between them about their intersection, while parallel lines give a translation by twice the perpendicular distance in a direction along the lines (see: https://mathinsight.org/reflection_line_2d).
A quick, reliable way to reflect a point across a slanted line is to “split” the point’s offset into along-the-line and perpendicular-to-the-line parts, then flip the perpendicular part. Concretely: for the line y = m x + b, a normal vector is n = (m, −1), and any point on the line will do-take Q = (0, b) for convenience. For your point P = (x0, y0), form v = P − Q. Project v onto the normal: proj_n(v) = [(v·n)/(n·n)] n, where n·n = m^2 + 1. Reflection just flips that normal component, so P_reflected = P − 2 proj_n(v). That’s the whole recipe: “subtract twice the projection onto the normal.” If you like a one-line version, plug in v and n to get P’ = (x0, y0) − 2[(m x0 − y0 + b)/(m^2 + 1)](m, −1). It sticks better (at least for me) if you remember the story: keep what’s parallel to the mirror, flip what’s perpendicular.
For two reflections: if the lines intersect, the result is a rotation about their intersection by an angle equal to twice the angle from the first line to the second (order matters for the sign). If the lines are parallel, the result is a translation, straight across the common perpendicular, by twice the distance between the lines; for y = m x + b1 then y = m x + b2, the translation vector is 2(b2 − b1)/(m^2 + 1) · (m, −1), pointing from the first line toward the second. These “double angle” and “double distance” facts come from the same flip-the-perpendicular idea: one flip changes the perpendicular component’s sign, the second flip either spins it by twice an angle (when the normals meet) or marches it forward by twice a gap (when the normals are parallel).
Want to try it on your old nemesis, say reflecting P = (2, 5) across y = x + 1, and we can check the numbers together?
Reliable method: write the line as A x + B y + C = 0 (for y = m x + b use A=m, B=-1, C=b), compute t = (A x0 + B y0 + C)/(A^2 + B^2), then reflect P=(x0,y0) to P’=(x0 − 2At, y0 − 2Bt)-it’s just “project to the line, then go the same distance again”; e.g., across y = x + 1, P=(3,2) gives t=1 and P’=(1,4). Two reflections in intersecting lines give a rotation about their intersection by twice the angle between the lines, while two in parallel lines give a translation by twice the perpendicular gap (see https://en.wikipedia.org/wiki/Reflection_(mathematics)).