I already accepted an answer for this, but I thought I'd post back because what I've learned since may be of help or interest to others.
After cleaning up the damaged coving and chipping off the wall plaster, it seemed more apparent (as some commenters suggested) that this coving was made in situ. In fact, it seems as though the plaster of the coving was the same continuous layer of finishing plaster as was on the wall. You can just about see below how it sits on the bonding coat.
After plenty of Googling I learned that the cove was probably not so much cast as sculpted using a running mould. There are some amazing videos of people doing this on YouTube. https://www.youtube.com/results?search_query=running+mould
I didn't fancy doing this in situ, but I found various videos of people using the same technique to create sections on the bench. Inspired specifically by this video I decided to give that a try.
Making the blade that scrapes the plaster is worth some attention. It was the most time consuming part and essential to get right. Here's a good video of that whole process: https://www.youtube.com/watch?v=rlH_vMc6yTE
I read that zinc is a good material for this, but I ended up using 0.5mm aluminium which is way too soft and flimsy for extended use. However, having very limited metal working tools this did allow me to cut a half-decent blade with tin snips and file into shape; finally smoothing down imperfections with 240 sand paper.
I won't post all the details of my ugly contraption, but I ended up with this.
It took 6 or 7 fairly thin coats of plaster of paris. The trickiest part was mixing small batches of plaster quickly between coats before the plaster could set on the mould. Plus also keeping the blade and running track clean between passes. I want an assistant next time as this was surprisingly exhausting.
Other notable details are that I added a layer of hessian at the back for strength. A curve of wire mesh would have been better (as in earlier video) but I was doing experiment no.1 on the cheap. Also, thanks to commenters in this related question, I used washing up liquid as a release agent.
A hollow back was created by putting a right angle bead at the back of the casing. This made the section lighter and used less of the fairly expensive plaster too.
Here's the final product out of the mould. I haven't glued it into place yet. That's for another day.