In 2003 I produced a small proposal to fit several two-bedroom houses onto a small site somewhere in Kent, UK. I wanted to design only one house with arbitrary window positions so the houses could be rotated and clustered on the triangular plot. This is it. It could be a row house with windows only on end walls or it could have windows in the side walls, end situation permitting. I liked the fact the plan could be shrunk yet still retain all its features.

It wasn’t horrible, but this is what I think can be improved.

• Apart from the downstairs w/c, the middle part of the plan is taken up solely by circulation. 27.8% is too much, even if it does allow one to see along the entire length of the house. I expect this excess width at the entrance point came about because I thought the roof geometry more pleasing.
• The upstairs bathroom is too large.
• I’m not convinced of the necessity for a downstairs w/c.
• The second bedroom is the same size as the living room.
• In a row house situation, the courtyard wouldn’t be large enough to function as a courtyard.
• The original proposal had communal garbage bins adjacent to the car parking area but there should be some provision for individual outdoor storage, if not necessarily for garbage bins.

This is the 2022 re-design.

• I overcame my misgivings about the roof geometry and circulation space shrunk by 10% to 17.6% accordingly. Short of introducing a ladder stair or artificially inflating the habitable area, I can’t see how it could be made any less.
• I toyed with entering the house from a re-entrant corner in front of the stair. This increased the habitable area organically, and also placed the entrance at a more useful position – as it is in many a terraced house – but it complicated the upper floor unnecessarily.
• I also toyed with the idea of having the houses repeated instead of mirroring pairs of them around the party wall separating the courtyards, but this meant the courtyard wall would need to be higher and would reduce light to the courtyard and the rooms opening onto it.

• Mirroring also meant a reduction in the amount of external wall area. All the reduction was for party walls so there was no loss of window wall area.
• The courtyard is now square and rooms opening onto it now have double the area of glazing. End walls need only have minimal window openings.
• Construction could be updated and panelized but I’ve kept it brick for now. Its still basically a K-Span house and, apart from the internal circulation, still isn’t world’s away from Ando Sumiyoshi/Azuma House. If you want to link two rooms per floor around a courtyard, then it’s difficult for it not to be. I’ve not gone with his peculiarly Japanese bathroom positioning but I have stacked the wet rooms.
• The general shape and slope of the roof remain unchanged and the roof still discharges to the courtyard and then via a drain to the street. However, parapets are additional cost and (in order to prevent the spread of fire) are best reserved for party walls only. Pending.

New Shikumen I

The defining characteristic of the Shanghai li-long typology is the internal courtyard existing in a row-house typology, and the defining characteristic of a shikumen neighbourhood is these row houses arranged in rows inside a block having perimeter retail. These rows are single-sided with the front of one facing the rear of another.

• The bathroom can have a window.
• A L-shaped kitchen is possible. The kitchen has a service access with storage and space for garbage bins.
• All rows are oriented north-south and so mirroring houses will not affect daylight quantity.
• The service area can hold two garbage bins and is also an alternative entrance.
• Next to the service entrance is a storage/utility area. It’s possible to have a lower level w/c, and with a window this time.

• I spent much time pondering the best configuration for the roof. I’ve kept parapets between houses to prevent fire spread and I increased the party wall projection between adjacent upper level windows for the same reason. Whether to keep the parapet walls flat or have them follow the inclination of the roof will be a tradeoff between economy of materials and economy of labour, assuming conventional construction.

• I assumed conventional construction, even knowing that the walls will probably not be cavity brick and the windows timber framed. I’ve kept the construction simple and avoided lintels. The only internal beam spans the corridor to support the roof at the bathroom end of the stairwell.
• In the end, I set the direction of the roof for maximum daylighting to the courtyards and to the streets.

All that needs to be done now is multiply them. I’ve made no provision for car parking. It’s in the street which will is about 6 metres wide – not unlike many a London mews house.

Unsurprisingly, the streetscape resembles 1920s Shanghai where this building and urban typology originated to solve chronic overcrowding. The premises are still valid. The space between rows could even be reduced to 1920’s standards if some alternate provision were made for vehicle access. The street with in the image above assumes two way traffic and cars parked on one side. The street width in the images below was most likely arrived at from daylighting and ventilation concerns rather than automobile access, and is probably minimal.

Narrow as the streets of this proposal are, they will still be more lively than the circa 1960 streets of an Arab city where courtyards are completely enclosed and windows rarely open directly onto streets for reasons of privacy as much as security.

Finally, the area delineated in the image below is approximately 13,100 sq.m. There are 2 x 8 x 5 dwellings in that area = 80 with an average occupancy of three persons = 240 people = 18,500 ppl/sq.km.

This is a population density greater than Seoul and roughly equal to that of Macau, but still less than Athens.

As ever, the point of these explorations is not to design a tiny house in isolation, but to design one so can be aggregated with many others without sacrificing any benefits of the layout.

• For a while, I thought the small piece of connecting roof above the stairs and corridor could be some kind of fully glazed orangery-type connection covering the stair and a bridge corridor. It’s a possibility. This space would get hot and cold, but not as much as Ando’s Sumiyoshi House.

Perhaps getting rid of the parapets and having a box gutter isn’t such a bad idea. Putting that box gutter not along the parapet but along the line of the upstairs corridor wall will keep the height of the butterfly roof low. Murcutt-esque curved sheet metal would be cool but is’nt going to happen. This is where I’ve left it for now. The box gutter is simple a normal gutter along the courtyard wall. It is open at both ends to prevent blocking and overflowing and to let rainwater drain directly to the street.

---

Shikumen (石库门)

New Squeeze