When I first read Rob's note just now I didn't understand at first what he meant, but I think I do now see the point he is making. With no vanishing points, there are in effect multiple horizons.
I am away from base at the moment, so don't have access to my PC. I was going to re-post here three photos (originally published in the MRJ article) that appeared earlier in this thread, which I think illustrate what Rob is saying, but I have not been able to copy and paste them. So readers may have to scroll back to page 2 of this thread to look at them, or get out MRJ 220 and 221.
The first of these shots is the photo at the head of the post of Saturday 23 March 2013
[printed on page 4 of MRJ 220], the second is the photo at the head of the post on Friday 29 March 2013
[printed on page 52 of MRJ 221], and the third photo is the fifth one in the 29.3.13 post
[printed on page 54 of MRJ 221]. Each of these three shots shows a variety of buildings turned at various angles to the viewer, but their roof lines (and other lines, such as door and window heads, etc.) are consistently horizontal. The buildings are ranged up a hillside, and so each building, or group of buildings, in effect has its own horizon. But the fact remains that there are no vanishing points on any of those multiple horizons.
If you look carefully, I had difficulty in keeping all the roof lines absolutely horizontal, because of my natural habit of drawing in perspective, but I think they are near enough level as makes no appreciable difference.
[Thanks, Keith, for the edit. Much appreciated.