91. South Bridge, Edinburgh

So, this was promised a couple of months ago. Like George IV Bridge, South Bridge is well disguised from above with only a short gap between buildings each side of the road. Said buildings project 3 stories and more above, but as can be seen here they also reach 3 stories below the road.

 South Bridge from Cowgate

South Bridge from Cowgate

Again, the length of street with the name “South Bridge” gives away the fact that this span over Cowgate is part of a viaduct crossing a wide valley. The far side in this image is set lower. It has windows below street level on the far side so it is presumably some form of corridor between properties each side of the Cowgate, though the one to the left as we look burnt down some time ago and what can be seen is obviously new build.

This is a view of a model on Sketchfab. (Not a photogrammetric model, this one is built up from primitives, note the faceted rendering of the arches.) It shows rather more spans than I remembered, and also the many “floors” in the spans, which I will return to below.

The arches are semi-circular. The Cowgate span looks about 9m (30ft), the remaining 16 are 6m span. The bridge has been widened on both sides.

I have looked at this bridge many times but was struck by details I didn’t remember. Maybe the light flooding in from the early evening sun highlighted things. I used the Sony HX90V that lives in my belt to take 100 photos, which Hamish built into a 3D model for me. All photography was from the footways in about 10 minutes.

Our own viewer lets us produce images like the one below. An orthographic view vertically upwards.


I used the camera systematically, altering the tilt then walking through taking a line of pictures so I got full overlap. I used the widest angle available (24mm at 35mm equivalent). For this job, more photos at a narrower angle would be advantageous. Still the image is good at this scale.

Enlarging a patch near the crown at the right shows that the resolution is limited. Even this, from photography completed safely in 10 minutes without closing the road, offers insights that could easily be missed in close inspection. At minimum, a model should be studied as part of the pre-survey work.


Our viewer allows us to put a horizontal plane through the model and this shows the structure to be in good shape generally. There is no mis-alignment between the two contours and no step where the extension on the right joins in.


Moving the cut closer to the crown creates rather more amplification in the contours.


The contours get closer together as the arch drops so here there is a definite hollow and step near the join.

The viewer renders the back side of surfaces as transparent, so we can easily produce an orthographic elevation, here looking North.


The flood of bright sun on the south side burnt out part of these images but it is a measure of the software and sensor combination that we got a good model for much of this. And here is where I started to see new stuff.


There is a pronounced run of steps in the face at the string course, right hand third point in this image. It is worthwhile taking a crop from the model (Left below) and comparing it with the original to see what is lost in this low density model.


The step line runs through the string course, as is emphasised by the lime runs below. In the photo, there is possibly a visible run going the other way to the left which I wouldn’t have noticed in this model. The step in the arch at the top of the two runs, though is visible enough and very wide.

I know from previous involvement that the rest of the viaduct has flat arch (about 6m span 1m rise) brick floors at various levels (shown in the model linked above). The spaces have been taken over by various people and some have gone in for demolition of the floors. I wonder whether the steps are evidence of damage resulting from careless and uneducated demolition generating very high thrusts.

The round hole there is surely a core but the rectangular openings must also be there for a reason. The pattern is random Are they drains? If so, why there?

The Tell-Tales pick out a longitudinal crack but there is extensive damage around it and much more thought is needed than appears to have been applied here. For a start (but I would say that) I would want Moiré Tell-Tales on the crack so they could be read from the ground and also recorded in video to look for live load movement. The loose and displaced stones are certainly an indicator of that and such damage will only get worse. Worthwhile treatment will be difficult and some serious investigation should precede any intervention. More importantly, in the first instance, nothing should be done that cannot be removed when it turns out not to work!


Looking at that area in the photographs is more revealing, but would I have seen it at all without the model to explore in peace, and would I have known where it is if I did?


Cropping and magnifying the photo shows much more detail (of course) and shows just how good a modern camera can be in these quite difficult circumstances.


I would really love the opportunity to look closer at this.


What I did do, next morning, was have a look on the road above. I didn’t get time to look at the West side of the road (where the crack is) But the east side doesn’t look ideal in this pic from my phone.


And finally. The connection to the extension at the east side is very interesting. I’m reasonably sure this is a join but the stitching must have been quite tricky to do, getting those big stones to slide in to the interlock. Looking at it now, I guess there would have been more stitches had the stones all been the right thickness.