Ok, so, for the first time I played with this. Kind of maddening. I found that if you do not use EXACTLY the endpoint of the trajectory curve as your "alternate origin" that you end up with lines that are not exactly coincident with the trajectory curve. And, you're stuck with the very first instance, though you can turn off any other. It's neat finally being able to do this, but it's got it's quirks for sure.
AND......what would it do on a helical or other 3D curve? Is it smart enough to know, on curvature, what side to be normal to? Towards the center or away from the center? Hmmmm.....