In polar coordinates, we can write:




We also know that $h=r^2\dot{\theta}$ is a constant. This can be shown by math or noted from the fact that gravity doesn't exert torque.

However, in polar coordinates we can write:

$\hat{r}= \cos\theta\space \hat{i}+\sin\theta\space\hat{j}$

$\hat{\theta}= -\sin\theta\space \hat{i}+\cos\theta\space\hat{j}$

Thus, taking derivatives with respect to $\theta$, we get:

$\frac{d\hat{r}}{d\theta}= -\sin\theta\space \hat{i}+\cos\theta\space\hat{j}=\hat{\theta}$

$\frac{d\hat{\theta}}{d\theta}= -\cos\theta\space \hat{i}-\sin\theta\space\hat{j}=-\hat{r}$


$\frac{d\hat{r}}{dt}=\frac{d\hat{r}}{d\theta}\frac{d\theta}{dt}= \dot{\theta}\space\hat{\theta}\tag{3}$

$\frac{d\hat{\theta}}{dt}=\frac{d\hat{\theta}}{d\theta}\frac{d\theta}{dt}= -\dot{\theta}\space\hat{r} \tag{4}$

So, reorganizing (2) as:


We substituting (4) in (1) we get:


Since h is constant we can integrate. Integrating $\frac{d^2\vec{r}}{dt^2}$ with respect to t, we can say:


Now, to find the tangential component of $\frac{d\vec{r}}{dt}$, we can take the dot product of $\frac{d\vec{r}}{dt}$ and $\hat{\theta}$.



My confusion is because what I have shown is that there is no dependence on $r(t)$ in the tangential component of velocity. If the radius is changing but velocity remains constant, then h must be changing. So how can angular momentum be conserved? Is this a contradiction? Is there a flaw in my argument somewhere?

Also, one last note: $$\frac{dv_T}{dt}=\frac{d}{dt}(\frac{GM}{h}+u\cos\theta)$$$$=u\sin\theta\,\dot{\theta}\tag{6}$$


Since $0=2\dot{r}\dot{\theta}+r\ddot{\theta}$ in (2),


Finally, combining 6 and 7:

$$\frac{dv_T}{dt}= u\sin\theta\,\dot{\theta}=-\dot{r}\dot{\theta}$$ $$u\sin\theta=-\dot{r}$$

This can't be right! Integrating gives a non-conic: $$r=u\cos\theta+C$$

  • $\begingroup$ One thing that is off is $\hat{\theta} \cdot \vec{u} = - u_x \sin \theta + u_y \cos \theta $. $\endgroup$
    – secavara
    Jul 17, 2020 at 18:31
  • $\begingroup$ @secavara True, except I don’t see how that could make the L/mr dependence that you’d expect for velocity $\endgroup$ Jul 17, 2020 at 18:34
  • $\begingroup$ I don't see a mayor conflict between $\frac{d \vec{r}}{d t} \cdot \hat{\theta} = \frac{G M}{h} - u_x \sin \theta + u_y \cos \theta = \frac{L}{m r} = \frac{h}{r}$ and the general parameterization for an ellipse relative to focus in link. $\endgroup$
    – secavara
    Jul 17, 2020 at 18:58
  • $\begingroup$ Or a general conic section, to be more precise: link. $\endgroup$
    – secavara
    Jul 17, 2020 at 19:09
  • $\begingroup$ You're right. There isn't a serious conflict. I still don't see how $\frac{d\vec{r}}{dt}\cdot\hat{\theta}$ is impacted by $r$. I'll edit my question... $\endgroup$ Jul 17, 2020 at 19:35

1 Answer 1


Got it!

The dependence is there, although not obvious.

Referring to $v_T=\frac{MG}{h}+u\cos\theta$, we can note that the path will be some sort of conic section, or at least something that can be described in polar coordinates: $$r=f(\theta)$$ $$\theta=f^{-1}(r)$$ Thus, $$v_T=\frac{MG}{h}+u\cos\, (f^{-1}(r))$$

There's the dependence for what it's worth!


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.