2

I follow a suggestion from a question on meta stackexchange to ask about actual, physical architecture here. I hope that is acceptable.

I am interested in the information and features which a knowledgeable observer, e.g. architect or real estate agent, would require to reach an informed estimation on the construction period of a building which is within his/her area of expertise.

conditions:

  • no information of the construction materials
  • no access to the interior of the buildings
  • only residential building
  • only ordinary buildings from the general building stock (no outstanding features)
  • allowed is information on the exterior, i.e. facade (from pictures or in situ inspection)
  • allowed is information on the topology of the area , e.g. surrounding buildings (again no information on the age), distance from city center ...

A little background to this question: I want to train a machine learning algorithm to recognize construction periods of buildings in order to use the information for building energy simulations. Here is a similar study.

Experiences from any geographic region are welcome.

closed as too broad by Brian Z, Denis de Bernardy, KorvinStarmast, José Carlos Santos, John Dallman May 23 at 18:23

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • Comments are not for extended discussion; this conversation has been moved to chat. – Mark C. Wallace May 30 at 10:41
  • The question is precise. I believe the OP knows that the task is difficult/challenging, and I believe that the OP knows there are other ways to do the task. This is not my field, but even as an outsider I know that (for example) the presence of a keystone establishes a maximum age for a building. The presence of flying buttresses or mansard roof are a strong indicator. I've been on enough architectural tours to know that there are architectural elements that are used to estimate the date of a house. I don't think 100% precision is the goal. – Mark C. Wallace May 30 at 10:45
  • The year funding is made available is an excellent predictor. – C Monsour May 30 at 16:37
1

Anyone can learn to do so to a degree.

As a teenager watching Dark Shadows (1966-1972) I saw that the fictional Collingwood estate at the fiction town of Collinsport, Maine, had two mansions. The present Collinwood Mansion and the abandoned Old House.

In episodes where a character traveled back in time to the 1790s, the newer Collingwood was new built, meaning that the Old House was even older.

But even as a teenager it was apparent to me that Collinwood Mansion looked like it was built a century or more after the 1790s, while the Old House looked like it should have been built at least 30 years after the 1790s.

The exteriors of Collingwood were filmed at Seaview Terrace or the Carey Mansion at Newport, Rhode island, which has a highly complex history since a large part of it was first built in Washington DC in 1907 and later moved to Newport and rebuilt there in 1923-25, incorporating Seaview, a house built in 1885. Some of the rooms were much older, brought from Europe to Washington and then to Newport.

https://en.wikipedia.org/wiki/Seaview_Terrace 1

The exteriors of the Old House - built in 1767 in fiction - were filmed at an appropriately abandoned mansion in Tarrytown, New York, near Washington Irving's mansion and near Lyndhurst, which was used as Collingwood in two Dark Shadows movies. The mansion used for the Old House was first built in the 1850s and probably had its columns added decades later.

http://www.hudsonvalleyruins.org/yasinsac/spratt/spratt.html2

https://darkshadows.fandom.com/wiki/The_Old_House3

So as you can see my teenage self could tell fairly accurately that the buildings used for exterior scenes of Collingwood and the Old House were much younger than the fictinal ages of those buildings.

No doubt an architectural historian can estimate the ages of buildings much better than I can, and some of the knowledge of architectural historians can be programmed into computers. Therefore, it seems to me that what you ask about could be done, but I do not know how well it could be done.

1

I'm not an architectural historian, although I have an interest in the subject, but I really do not think this is feasible, at least without limiting your area of interest considerably.

To take just one vernacular form of architecture, wattle and daub. This is a widespread technique, with many variations in construction, materials and wide geographical spread. How an algorithm could distinguish between a medieval English house and a similar construction in, as Wiki Commons illustrates, Bulgaria, frankly defeats me.

Furthermore, architectural historians rely, not merely on stylistic conventions but techniques of roof construction, material, joints, wall plates - all of which require access to the interior, often extremely difficult to physically get at. Then, there is dendrochronology - when were the timbers felled? At the time of original build, or during a later remodeling /repair?

I am not a computer programmer, but to construct (pun intended) an algorithm to do all this from external photographs appears to me quite impossible. Sorry!

Edit

I apologise, I did not pick up on the energy implications. However, this stuccoed house is a straw-bale construction, an excellent insulator , but I doubt this would be obvious from an external photograph.

Exterior view of straw bale library in Mattawa, Washington taken in 2008
Exterior view of straw bale library in Mattawa, Washington taken in 2008

  • I'm not entirely sure that the distinction between a Bulgarian wattle and daub and an English wattle and daub is significant for energy simulations. – Mark C. Wallace May 23 at 15:00
  • 1
    @MarkCWallace - presumably the materials used might be. – TheHonRose May 23 at 15:02
  • 2
    How to distinguish this is easier with rough location known. Then it's not just the pixels showing the walls, but all the other elements comibined into a statistic that would enable one probability estimate. For much of 20th century, I guess results of a properly trained algo could be pretty amazing in accuracy. I've often seen untrained people do just these guesses around their own locality, just based on experience and feeling (for style), come out quite close to the real date. (The energy angle might be quite off in general, though. But ultimate aim of OP is a lesser concern) – LangLangC May 23 at 15:28

Not the answer you're looking for? Browse other questions tagged or ask your own question.