Thanks for the heads up about the NewPing library Dave - I missed that, seems really handy! Definitely could use the median reading feature as well for when the readings need to be more stable.
The sensors were on a triangular prism tower (so pointing in different directions) and a good distant apart from each other so I don't think they were interfering too much - I found the problem when only one tower was powered up as well so it wouldn't be the other towers either.
Good to know that can be a problem as well, I tried to account for interference but could probably be a bit more careful about that.
For now I guess I'll try adding a timeout as I need a quick fix for a temporary version of this installation on a film shoot in a couple weeks, and I don't know if I'll have time to rewrite the sketch with the NewPing library just yet.
Thanks again both of you for your input.