The angle transformation under Lorentz transformation:

$$ \cos\theta = \frac{\cos\theta' + \beta}{1 + \beta\cos\theta},\quad \sin\theta = \frac{\sin\theta'}{\gamma^2(1 + \beta\cos\theta')} $$

is normally derived from the velocity transformation:

$$ u_\parallel = \frac{u_\parallel' + v}{1+(v/c^2)u_\parallel'},\quad u_\perp = \frac{u_\perp/\gamma}{1+(v/c^2)u_\parallel'} $$

and then

$$ \tan\theta = \frac{u_\perp}{u_\parallel} = \frac{u_\perp'/\gamma}{u_\parallel' + v} = \frac{u'\sin\theta'}{\gamma(u'\cos\theta + v)} $$

letting $u' = c$ and get:

$$ \tan\theta = \frac{\sin\theta'}{\gamma(\cos\theta'+\beta)} $$

I am wondering why we don't start with the length transformation:

$$ x = \gamma(x'+\beta ct'),\quad y = y' $$


$$ \tan\theta = \frac{y}{x} = \frac{y'}{\gamma(x'+\beta ct')} $$

which leads to an annoying $ct'$ here

Can someone tell me why we must use velocity transformation and not length transformation to get the angle transformation?

  • $\begingroup$ Why not start with "time transformation"? Velocity is not any kind of "associated quantity" with length. $\endgroup$
    – m4r35n357
    Oct 29, 2023 at 11:05
  • $\begingroup$ time transformation is associated with length $\endgroup$ Oct 30, 2023 at 5:47

1 Answer 1


I think, since you are only considering 2D (as oposed to 3D), i.e. a component of the velocity that is perpendicular to the Lorentz transformation and another that is parallel with it, you are allowed to say that $$x'=r'\cos\theta'\\y'=r'\sin\theta'$$ for some distance $r'$. The latter distance, then, is going to be the distance that satisfies $ct'=r'$, and therefore you will have the Eq. in your original post $$\tan\theta=\frac{\sin\theta'}{\gamma(\cos\theta'+\beta)}$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.