Friday, September 2, 2011

Enter the Octagon (III)

In the foregoing post in this series, I looked at Orson Squire Fowler's work in the mid-1800s to popularize Octagonal houses. I spent some time demonstrating the inherent efficiency, when measuring the amount of floor space achieved for a given amount of liner feet of wall, of the octagon over the square and rectangular plans. Personally, I found this basic number-crunch quite eye-opening, as it has shifted my ideas about the forms houses might - or ought to - take.

Personally, I've always tended to like rectangular houses,  particularly ones with ells, wings, u-shapes, stepped patterns, etc. I find them visually interesting for one thing. I like Japanese houses with staggered patterns, what they term the 'goose-formation' pattern, along with New England connected farm buildings of a similar configuration, a pattern here christened the 'big house little house back house barn' pattern. Now those forms of buildings, historically speaking, evolved over time. People needed more space, or didn't see the sense in cold weather in having to walk outside of the house to get to another building, so they added on to one building with another, in a graduated fashion, as budget permitted. When you're adding on to space, it is much easier to tack one building on to another, make a hole through the adjoined walls to connect the spaces and stitch the colliding roofs up in some way - something not always especially elegantly done from what I have seen. This is the more obvious approach to adding space, as one doesn't have to open up the roof on the existing house and modifications to both structures are minimal.

In Japan, the sense of architectural space is such that there is a definite preference for spaces in which one 'discovers' new views and scenes by turning the corner, each room connected to another and yet potentially very different in feel or theme. Like pearls on a string. Or like stations on a subway line -  the dark route between individual stations is not so critical, it is the stations themselves that are the interesting part.

It is one thing when building have evolved over time into forms like that, and it is quite another to construct them from new in such a pattern. There's more than a little of that going on around here - near where I live there are entire subdivisions of new houses with attached pseudo back-houses and pseudo-barns. Not a lot of small-holder farming going on there I suspect. Fowler decried especially buildings constructed on patterns of jogs or wings, as these manifest all the faults of long rectangular buildings (i.e., a lot of wall for very little enclosed floor space), but even to a greater degree.

Consider the following situation in which you place together two floor areas, one main space at 30' x 45', and the other, a 'back house' at 24' x12':

This is a fairly common pattern in New England, the back house being the new kitchen, and invariably later added onto until the overall plan becomes 'T'-form'. Note the portions of the inward jogs, where I have labeled the walls as 'a', 'b', 'c', and 'd'.  Adding up our square footage, we have (30 x 45) + (12 x 24), a total of 1638 square feet, and we needed 174' of wall to accomplish that.

Now, keeping in mind those labeled wall sections,  'a', 'b', 'c', and 'd', we could alter the configuration, and not add one single bit of extra wall, by moving them as follows:

I left dotted lines in the drawing to show the original plan. We now have a plan that is getting closer to a square in overall configuration. We have also eliminated a fair amount of cutting work, if you consider the three corners required at each end to accommodate the inward jogs have now been replaced by a single corner. Square footage in this plan is 45 x 42 = 1890, a 15% increase. Wall length is identical.

So building from new in a plan with jogs, ells, and wings, makes less sense economically in terms of outlay for materials and amount of cutting required. In a warm climate, I would say that long stretched out buildings of a single floor make sense if the goal is to maximize ventilation and catch breezes. In a hot climate, nobody is going to want a bedroom on an upper floor under the roof. A long stretched out house, as it has more wall area in relation to floor area, allows for a greater number of wall openings to allow that breeze in and through.  A skinny stretched-out house makes sense for such a climate.

In a cold climate, more external surface area only adds to the heating bill. It makes much better sense in a cold climate to have a condensed floor plan, in the form of a short rectangle, a square, or a polygon, and to have more than one floor, so that the heating of the first floor can contribute toward the heating of the upper floors simply on the basis of warm air rising. Wall penetrations should be minimized in such a house, especially on the northern side of the structure, or on sides regularly exposed to cold winds.

An octagonal house, in terms of how it faces the wind, is much less of a block than a rectilinear house:

Also, in terms of the structure resisting the racking effects on the frame that come from the wind pushing against a wall, consider that the square/rectangular timber framed house would employ diagonal horizontal braces at the interior corners, whereas with an octagonal house and entire wall is oriented on a 45˚ making it stable, on the basic of the geometry to the same loads. It's like the advantage of a hip roof over a gable roof in terms of the inherent triangulation gained.

Fowler felt that a 3-story octagon house was ideal in such a climate, as the bedrooms could be placed on the second floor so as to be in that happy medium between too cold and too hot. Having slept in houses on the top floor under the roof here in New England, I can attest to the fact that such spaces are not conducive to getting much sleep in the heat of summer.

There are a few websites out there which catalog the many octagonal houses in North America. I've looked at dozens of examples, along with those put forth in Fowler's book, and frankly I don't tend to find many of them especially attractive in terms of their appearance from the street. Part of that response might be simply due to the novelty of the structural form, and now that I see and better understand the virtues of the form I am beginning to view the octagonal house more favorably. I think the least attractive portion of these houses, coming from my perspective as admirer of Japanese traditional vernacular architecture, is that the roof doesn't look quite right most of he time. I especially dislike the commonly fitted central cupola, which to my eyes looks an awful lot like a lighthouse:

The above example is the Hyde Octagon in Mumford New York.

Another famous example of an spectacularly large octagonal house is in Natchez:

I really find the domes and central towers a bit much in these designs. For some reason I like towers on the perimeter of a structure, not so much in the middle.

But, mostly it is the roofs overall I don't care for, with one exception. The roof that suits this form, I do believe, is the mansard:

Here's another example from Massachusetts - remove that cupola on top and it would be even sweeter:

I was surprised to find, upon a little digging around, that there are companies in Japan touting the virtues of the octagonal house plan. Here's an example of a newer Japanese house built on an octagonal plan:

Nothing especially exciting about that house, in my view.

Octagonal structures do have a fairly long history in Japan, associated to Buddhism  - perhaps because of the 'eight-fold- path that forms a central tenet in most schools of Buddhism. Here's a few examples of octagonal temple structures, starting with one of the most famous, the Yumedono at Hōryū-ji (法隆寺) in Nara:


This one is Buddhist in origin, though located within the precincts of a shrine on Enoshima, an island in Sagami Bay:

Another example - hope this prolonged set of examples isn't too boring:


This one has an intriguing upper roof:

Let's not forget China:

Now, I happen to be one of those people who thinks that temples do not make good archetypes for building residential architecture, but I like looking at those buildings all the same.

A lot of people wonder what the interior arrangement of an octagonal house would be like, and what advantages, if any, such a plan confers. In Fowler's book, a number of different floor plans are proffered and explained in detail. Here's one:

And another:
In fact, if you do an online search for 'octagonal floor plan', you'll find hundreds of examples with much variety in approaches to layout. There are as many ways to slice up the pie, as it were, as with any rectangular/square house's floor plan.

One Japanese site advocating for octagonal homes had some great graphics which I have shamelessly pilfered. Hopefully by providing a link they won't sue me!

Comparing an office or study room in an octagonal house with a room in house with 90˚ corners, same square footage in each, we can see the following:

In the square-cornered room, much of the space at the corner is poorly usable, access to one of the bookcases is constrained, and if a view is desired, much of it is obscured:

In the octagonal house room, there is much more desk space, more room for the bookcases, and a more panoramic view could be had with additional windows:

Seems like a better place to hang out - the guy seems pleased. Yay!

Fowler also mentions the dead space one obtains with the interior 90˚ corners in buildings. He argues that the octagon's interior partitioning would allow for nearly every room to have a triangular closet, which he states are more accessible and useful spaces than one gets with rectilinear closets, which of course have those same dead spaces in their interior corners. He further lays out the benefits of a more condensed centralized plan, as it shortens the distance between places in a house making it less tiring to get from one place to another.

I think all in all that Fowler's An Octagon House for All has much of value to contribute to how we think about designing houses. Whenever we build, we need to think of site first. An octagon house can be placed extremely well on a variety of lot shapes, even triangles which are quite awkward places in which to put a rectangular or square house. Site also relates to prevailing winds and solar orientation. An octagon house would tend to present a flat wall surface to several positions as the sun arcs across the sky, making it a good choice for passive solar. The octagon house lets the wind pass by more easily than a house with square corners. Then we can think of building for time-  is an octagonal house easily modifiable, can it be added on to? Absolutely. And then, we can consider the usual matters of cost and square footage, where an octagon house is clearly superior in terms of efficiency of material use.

All that said, I have to express a certain amount of apprehension at the idea of framing a building with 45˚/135˚ turns, and 22.5˚/67.5˚ miters to deal with. While it isn't challenging carpentry particularly, after working on a deck for a local carpentry company a month or so back, I noticed that the previous carpenter had managed to configure the octagonal portion of the deck with a 41˚-49˚-49˚-41˚ bump out (that's one of the reasons why he was fired I guess), so maybe octagonal buildings are asking for trouble? Even square and plumb seems to be more and more of a rarity these days, so finding carpenters to frame, say, a roof on an octagonal house might be hitting 'tilt'? I hope not. It's not that bad is it?

My take on the octagon house is that it is a very good shape to build with, but it doesn't mean that from here on out it is give me octagons or give me death - look out Chuck Norris!. It means that when considering the matrix of site, climate, budget, and value for money, I now have a new lens to click into place. Octagonal houses make sense in a lot of cases, and I would now tend to argue against long skinny houses in colder climates. I thank Orson Fowler for opening my eyes to that.


  1. there's a bunch in Michigan :

  2. It's probably worth noting that the efficiency gains claimed for an octagonal structure are met or exceeded with a geodesic dome. Here in south Georgia, heating efficiency is not that big a deal, reducing heat gain is.

    I do however, prefer straight walls -- easier to hang pictures if nothing else.

    JG Gerth

  3. JG Gerth,

    yes, quite right in regards to geodesic domes, however they are otherwise poor in a number of respects - please see the comment section from post II in this series:


    Thanks for your comment!


  4. Chris

    The roof on the cover of Fowler's book is too little, too low. I'm not a big fan of the mansard, but its a step in the right direction. The temple roofs are just gorgeous. If a man's home is his castle, couldn't it be his temple, too?


  5. Tom,

    I would look to the 'Greek Revival' houses which dot the landscape of the US, in answer to your question...

    Thanks for your comment!


  6. Interesting reading, as usual. I have a log barn which is a pentagon. Each side is approx. 20' long when measured from the outside. The dividers for horse stalls come out from the five "corners" towards the middle, making each of the four stalls and one front entrance area pie shaped. (although not quite meeting in the middle, because that is the center walkway) For awhile I took down the dividers and used the barn as my workshop. Surprisingly, although not large I would get lost as to what section I was in, or what way the front door was, I think because of the angles of the outer walls. Placement of woodworking tools was also awkward, but it's a beautiful building (I can email you a picture if you're interested)

  7. Julie,

    thanks for posting. I would think that pentagonal structures generally are quite rare, and I have never come across a pentagonal barn. I would be delighted to see a picture if you have a moment. Thank you for offering!

    I think that for woodworking, given the machine orientations required for the efficient movement of materials through the space, that square and rectangular buildings might make the most sense, though I suspect Fowler would still argue for the octagon.Most of what i have read on shop design discusses the matter in terms of the type of materials one works with and how to move them from their unmodified state to the finished product in a smooth path of movement. This sort of thing might not be particularly relevant to the one-person artisan studio of course.


  8. Sorry, I don't buy the more efficient office diagram - the desk is almost unuseably narrow and by switching the desk and the shelves, rotating it 90º and putting the desk with its back against against the right wall you end up with a similarly useful space.. and another window could be put in the right wall for that panoramic view.. Just saying like..

  9. Adam,

    thanks for your comment - I understand that the picture might not be the most convincing demonstration of how octagonal interior spaces may be used. I think the point of that diagram, looking back to the Japanese site I pinched the picture from, was not well represented by my writing. In that section they are talking about the advantages in a space which combines right angled corner and wider angled corners in terms of flexibility in placing things. For showers, for instance, they mention it is easier to clean an interior 135˚ face than a 90˚ corner.

    The room space they were illustrating there is meant to be a den, and they mention that the long wide desk allows two school students to use it at the same tome, and for material to be laid out over a long surface. Those are advantages in certain contexts.

    Square interior corners do tend to narrow the possibilities of use I tend to think, and the point about interior 90˚ corners being largely dead space is a valid one. A desk placed into a corner is less accessible - only two sides are usable. It's better if it were in the middle of a wall. In the octagon space shown, you could put that desk easily along the long wall and access three sides.

    Of course, rooms can be configured in various ways. Looking at plans for octagon houses I note dozens and dozens of variations. I guess the point is that an octagonal interior space does not suffer from any clear shortcomings as compared to a rectilinear interior space - and offers certain benefits. Keep in mind that the interior of an octagonal space can be partitioned so as to have several rectangular or square rooms as well, if a room with 90˚ interior corners is what you want.


  10. I recall reading that octagonal outline require more difficulty if ever one wants to expand the dimensions. The perfect octagon, therefore, is loss whenever a new room is added on. Yet is two parallel side are expanded, do we not simply end up with a basilica layout? Is not the octagon house simply a series of giant bay windows seen from 8 different perspectives?

  11. Potomacker,

    thanks for your comment and I think that's a good point about any future expansion of an octagonal building - the shape is already a way of enclosing a lot of floor area for a minimal amount of wall, but should someone later want to expand the space they either can add a floor, if that is an option, or have to expand in ways similar to conventionally-shaped buildings, with ells, wings, bump outs, etc., all of which move away from the 'perfect octagon'. I think the beauty of the shape though lies less with the overall geometric form than what it means in terms of efficient use of materials.

    As for whether the octagon house is simply a series of giant bay windows from eight perspectives, well, why would anyone put identical windows on every face of a floor? You need a door somewhere, and the north side of the building (or wind -exposed side if there is a steady breeze from some direction) should have minimized glass/openings so I don't see every face of the building looking the same.


    1. Chris:
      I live in an octagon house, and wouldn't have it any other way. I have thought of making a one-story sun room off the back (the house is two-stories). I am guessing that would alter somewhat the "perfect octagon", although the addition would angle in so as to conform to the octagon shape. I do not share your dislike of cupolas and wish my house had one, although, I do agree that the mansard roof is more suitable to an octagon house. I could find so many uses for a cupola: meditation room, art room, small library, or just to have a magnificent view of the town. The down side of a cupola is that unless the windows are really tight and double paned, and the room itself is insulated, it would lose a lot of heat in the winter time. Still, I would find it impractical to try to add one on my house, as well as historically untrue to how my house was built in 1860. I won't dispute something that, in the end, is a matter of taste, but, no, I go for those cupolas, even if I can't add one to my house.

    2. While I like the light from a cupola or clerestory, I think it is a poor choice for a cold climate for the reason you mentioned: heat loss, in the very portion of the structure that should have the highest thermal efficiency. And you're right otherwise, it is largely a matter of taste.



All comments are moderated:

Anonymous comments will be deleted

Spam will be deleted

Comments containing links unrelated to blog content will be deleted