Different pieces of a building’s structure have different importance. Columns and lateral bracing, for example, are more important than ordinary floor beams or portions of floor slab. If the former fail, the building as a whole fails, while if the latter fails the collapse will be local and will not threaten the overall safety. A local failure can still kill many people, so it’s not something to take lightly, but there’s a real different between a building collapsing and 200 square feet of floor collapsing. In a very broad sense, floor beams are secondary to the columns, girders, and bracing elements in a frame.
If we take all of the steel or concrete elements that support usable floor area as “primary structure” then secondary structure consists of the connections of other pieces of the building: connections for facade elements, hangers for mechanical equipment, and so on. These are pieces of the building, usually steel, that are serving the structural function of transferring load, but which have no meaning in themselves: they are there to enable something else. A lot of the structure I’ve pointed out in the last few days from the pages of Architectural Terra Cotta is secondary structure of this type. The illustration above, the last plate in the book, shows several types.
Starting with the steel beams – the I beams and channels – we already have a classification problem. Some of the beams shown here and in the other plates are part of the primary structure, usually as spandrel beams. Some are secondary structure, installed as part of a cornice (and running to outriggers or brackets rather than building columns) or some other architectural element. My definition from the send paragraph still works: the spandrel beams are holding up floor area and so qualify as primary structure, while the cornice beams are only holding the cornice and so could be removed without affecting the floor area. All are part of structure because all are holding load to resist gravity and wind, and all could cause damage, injury, and death if they failed.
The steel angles are a mix: some are lintels and therefore more or less primary structure (not holding floor area, but making windows possible) and some are pieces installed only to hold decorative terra cottas: outriggers and small ornament-only beams.
Then there are the pure terra-cotta sports: the clips (bent plates), cramps (more bent plates) straight bolts, hooked (J) bolts, plate washers, and straight rods. The rods are the attachments for all hung terra cotta: they pass through holes in the hidden webs on the back sides of the veneer terra cotta blocks. The cramps are the attachments for vertical terra cotta veneer, and attach into slots adjacent to bed joints. It’s complicated, but as this plate shows it really is a system. Using those pieces you can work up support for pretty much any shape you can dream of in terra cotta.
I find it useful to think of this as layers of abstraction. The building frame at its purest is a three-dimensional cartesian grid in steel, but one that is related to the basic architectural geometry of the building. The secondary channels and angles provide a steel frame that relates to the geometry of the ornament, and then the clips and bolts provide connections from the terra cotta to the secondary frame. This idea is not limited to terra cotta: a little over a year ago I described the exact same three layers of abstraction in the metal skin of the Statue of Liberty. Maybe that’s the minimum amount of structure needed to link two unrelated geometries: one layer close to one geometry, one close to the other, and one in-between.
