Saturday, August 21, 2010

This One Rings a Bell (7)

Welcome to post 7 in this design and build thread for a Japanese bell tower, or shōrō. When I last left off, I had completed the eave build up for the lower roof, which consisted of a square-section kaya-oi, a rectangular uragō above that, in turn surmounted by the upper piece, the kami-uragō. I had completed all that, yet, as things turned out, I was not in fact done. Design and re-design was the reality, and as I mentioned in the last post there were a few things to figure out with that eave edge yet before I could proceed with the structural upper roof.

Some of you may be wondering why there are all these pieces at the eave edge. After all, in Western carpentry, it is typical to overlay the rafters directly with planking (or plywood in modern times) and then apply the roofing to that substrate. The eave edge is thin, to say the least, unless a covering fascia is applied and a gutter (which blocks the view of the fascia, does it not?).

While there are several important differences architecturally between Japanese and the western vernacular architecture, one of the most primary is that of where the emphasis is placed. In Western architecture, the house is essentially a decorated box, as a certain F.L. Wright said some years back. That elaboration and ornament of the building revolves around the fenestration - how the facade of the building is configured, how the windows are placed/aligned, how the doors are visually framed, what pattern of window muntins are used, what paint or siding is to be specified, etcetera. The roof, particularly since the advent of modernism, has become reduced greatly in emphasis. It's just a lid on a pot so to speak.

As I understand it, the disillusionment that followed the first World War in Europe led to a certain ideological perspective in the influential design houses of the day. This perspective rejected hierarchy and authoritarian social structures in favor of more socialist ideas, design for the "common man" (= cheap) and so forth. The eave on the roof was considered by these designers to be symbolic of the crown, so, for that reason, the eave was eliminated as a design idea and the flat plane roof without eave came to be the favored choice. A case of ideology trumping reason once again. The eave helps the rain water avoid running down the walls after all. The roof does need to keep the weather out, or at least one would think so.

This affectation lingers on in the west, even in non-modern style structures. The roof is largely an irrelevant architectural feature. The eave, if there is one invariably features a monotonous white perforated metal soffit, and the only function to having a thicker eave edge might be for mounting gutters. Yes, I simplify the argument a bit here, for brevity's sake.

In Japan, the roof is everything, whether we are looking at a thatch-roofed farmhouse (minka), or an imposing temple. The word for roof in Japanese, ya-ne, (屋根), literally means the root () of the house (). In the Japanese traditional perspective, there can be no house without a roof. That may sound obvious enough, but it goes further than mere practicalities of keeping the rain off the insides of that box.

The various elements added to the eave edge of a Japanese traditional roof are rich in diversity of form and arrangement. Depending upon roof covering material (tile, thatch, wood shingles, asphalt shingles, bark shingles, slate, metal, etc.), and how the under eave space is to be decorated, the eave edge will vary in treatment. The pile up of structural elements at the edge of the roof serve structurally to tie the rafters together at their ends, provide a space between exposed and hidden roof components into which cantilevers or other eave strengthening elements may be easily added, and convey a certain aesthetic value in and of themselves. A thicker roof edge suggests a more robust and solid roof, even if the thickness is confined only to the last 20~50 cm of the eave edge and the main body of the roof has a much thinner actual build up of roof material. That edge thickness may be acquired by stacking shingles many layers deep at the edge, or by piling up layers of wood, as I have done in this bell tower, or by a combination of the two.

One point common to both tiled and metal shingled roofs, particularly on temples, is the use of an inclined board atop the build up of eave edge fascia. This inclined board is pitched differently to the lower support pieces like the kaya-oi or uragō. It's called a fuki-ji (葺地) - here's a view of a roof rafter with these parts laid on top:

The rafter is on bottom in orange, the kaya-oi is in green above it, and uragō is on top of that piece in a light brown color. The fukiji is the piece in red at the very top, along with the associated thinner support board below it in yellow. Notice how the front face of the fukiji is tilted away from the front faces of both the uragō and kaya-oi. The tilt of the fukiji, in fact, is 90˚ to the line of the upper roof rafters (not pictured). The fukiji, associated to the upper roof, serves as a transition piece between the lower roof, which includes the uragō and kaya-oi.

Here's how the fukiji is used in an application of copper roof shingling:

See how the roof shingles are continued down the face of the fukiji from the roof surface above. In fact, the copper continued right around the underside of the fukiji support board (in yellow in the first illustration) and ends at the junction with the uragō. Thus, any water that runs over the roof edge will travel down a protected surface and will drip off without having any opportunity to touch the woodwork. Another advantage to the fukiji is that it allows the field rafters to connect to it with simple square end cuts - with the kami-uragō employed instead of fukiji, the rafters need to be tapered out to a thin end to fit cleanly. It is also simpler to terminate the roof substrate onto a fukiji than onto an uragō or similar.

Here's another example of the fukiji treatment:

The fukiji transition can also be effected with shingles of another sort:

On a tile roof, the fukiji is typically scalloped out to receive the individual roof tiles (this one is only slightly tilted forward though):

So, that's the fukiji element. Looks straightforward enough, simple even, but, things ain't what they seem, particularly in the case of a roof with a curved eave edge towards the hip rafters. with those elements in the eave edge that tie to the lower roof, their surfaces are geometrically in plane with the rafters; thus, as the rafters rise, the fascia elements rise perpendicular to the rafter as well:

Now figuring out the curves and intersections for those parts is entertaining enough, but with the fukiji, given the fact that its face is along a different slope than the elements below it, we have the following effect when the roof edge curves up:

While the kaya-oi and uragō rise up in unison, the fukiji wants to fly south for the winter: as the roof rises, the fukiji moves further and further away from the pieces below, geometrically-speaking.

Here's the graphical depiction of the effect of a certain amount of rise on those elements:

Notice what happens to the distance 'A' in the drawing, which elongates to 'B' at the end of the rise. The point being, looking from below at the eave edge, one can observe the parallelism of the reveal at 'A'. If the fukiji is simply lifted up, then the gap increases, which is to say that parallelism is destroyed. This doesn't look so good - like the upper layer was put on crookedly to the layer below. The Japanese solution to the 'fukiji problem', as I term it, is to bend the fukiji piece back inward so as to maintain parallelism with the uragō's front edge .

Welcome to the curious and often very counter-intuitive world of compound layout my friends.

I first became aware of the fukiji issue on the Ellison project in 2003 or so when my co-worker Walter asked me about a section of a textbook, but one page, that mentions the matter. It has occupied my mind ever since as one of the curious layout issues that I didn't fully comprehend, and now that I have at last the opportunity to do a structure with a copper shingled and strongly curvilinear roof form I can at last face the demon.

It's been a struggle for the past few days getting to grips with the fukiji and learning how to correct the distortion. Here's a shot from the drawing showing the development of the 3D part from the 2D developed plan:

Here's a glance at the other side of the piece, where it is perhaps apparent how the front surface of the fukiji is somewhat curvilinear, just like the undersurface:

So, the fukiji is an example of the dreaded double curvilinear work, and it was a slice of delight let me tell you. I still have a few hairs left after the pulling and cursing! Actually, it wasn't so bad though I did draw it 4 times before I had the configuration and proportions I wanted.

Back in the bell tower, the kami-uragō was ditched and the new fukiji installed in its place - let's see how that turned out.

Before (with kami-uragō) on top:

After (with fukiji):

One very slight change engendered with the fukiji is hardly noticeable - a 0.5 cm extension of the hip nose.

Here's the look from up over the edge, staring at the innards:

Sighting right along in parallel with the face of the fukiji, a view no one will be able to readily have, the slight re-curve inwards is evident:

And from below, a seamless parallel line between fukiji and the uragō below:

Right then, the lower roof is officially done and I am moving on to the structural roof above, where the pitch will be 7.5/10. Danger, curves ahead.

Thanks for checking out the Carpentry Way today. Take care! --> on to post 8


  1. Chris,

    thanks so much for letting us look over your shoulder as you are designing the bell tower. Its hypnotizing, like those slow motion movies of a flower blooming. Can't wait to see how the structural roof ties in!


  2. Wow! this is awesome. I never thought that it would be possible to do this. Also, that it's cool to see pictures of carpentry work. This is really cool! Thanks.

  3. Tom, glad you are finding the process enjoyable to follow!

    Carpentry Masters, yes, it's very possible to do all this, just takes time and patience. I hope you will keep following along with the build.


  4. Hi Chris,

    I have been trying to find solid information regarding the projection angle of the kayaoi in respect to the kioi based on a roof structure that uses a two tier rafter system (base and flying rafters). As mentioned in this post, the fukiji is at a steeper projection angle than the kayaoi, however I have mixed information regarding the kioi based on multiple diagrams and cross sections I have dug up over the internet. Some say the projection of the kioi is the same as the base rafter tip and different in projection to the kayaoi. While some other diagrams say that the projection of the kioi is the same as the kayaoi.

    I suppose, however, that the projection of the kioi is the same as the kayaoi, otherwise another "fukiji problem" emerges. However, if that is the case, how do we determine the projection angle of both these pieces? Does it follow the angle of the base rafter tip or does it follow the angle of the flying rafter tip? Or am I overlooking something?

    Thanks for the great topics posted.


    1. Frank,

      there are various proportioning systems out there for the projections of the kioi and kayaoi line. That is one matter. The other concerns the slope of each rafter tier. Almost invariably, the flying rafters are at a slacker angle than the base rafters, and this does indeed produce another problem similar to the fukiji issue. This problem revolves around rafter spacing, perticualrly where the kioi meets the side of the hip rafter - the dreaded 'end of logic rafter' or ronji-daruki. This is a layout problem not solved in Japan until the 1800s.


    2. Hi Chris,

      Thanks for the quick reply.

      Some more questions. I noticed that in some tokyou bracket mockups and diagrams I observed, the kayaoi projection angle matches that of the tip of the flying rafter angle (assuming 2 tier rafter layout). Even the first diagram (cross-section) posted above shows this. In respect, the kioi has its projection angle matching the projection angle of the tip of the base rafters. Is this how the carpenters determine the angle of projection for these pieces? And if that's the case, the projection angles of the kayaoi and kioi are fated to be different. So does that mean that the kioi must be bent backwards, like the fukiji, such that it follows in "line" with the kayaoi?

      Regarding the ronji-daruki, it is something new to me. I took a step back to some of my older models and found that the ronji-daruki is indeed not quite done correctly. My remedy to the problem will be to widen the hip rafters slightly.

      Thanks for sharing your experiences and knowledge. For someone who doesn't read too much Japanese, I found out that I got most of my questions solved by reading your articles as well as discovering new ideas and terminology along the way.


    3. Frank,

      thanks for the follow up and excuse my delayed response. The connection between the angle at which ki-oi and kaya-oi rise is not related to the angle at which the rafter tails are cut. Often they are the same angle, but rafter tails are not always cut 90˚ to their slope. There are also cases where both base rafters and flying rafters are themselves curved along their length

      The fukiji is a piece of wood resting atop a surface which is curving upwards, relative to the slope of the pieces below, however the fukiji leans at a different slope and -- well, the issue was explained clearly above. With a ki-oi, it is not resting upon a surface with a different slope. It's curve is set relative to the eave curvature, as is the kaya-oi. So the same problem does not occur. The upper surface of the ki-oi need not be parallel to the lower surface - it depends upon how the flying rafters are framed. The ki-oi and kayaoi are also not close enough together that any divergence in their lines would have to be quite pronounced to be noticeable.

      And I do not believe that making the hip wide to accommodate the ron-ji daruki problem will solve it, as you simply push the problem along to the kaya-oi. The spacing of the rafters on both tiers should be the same, that is, aligned to one another.

      Two tier roof work is quite complex, especially when curves are involved. I have a fairly good understanding of the issues after a lot of study of Japanese texts, but tackling one of these eave structures would be an ambitious project. And I read Japanese decently enough to understand the layout texts well. So, I'm thinking that gleaning material off the internet - I've done similar searches myself - is not going to provide you with a complete picture. At least that is how it seems to me, however, have at it.


  5. Hi Chris,

    That clears things up a lot. Yes, double eaves is an ambitious project with many calculations involved. For myself, I have done a lot of 3D model sanmons, shoros, kyoto style kabe walls, tous (pagoda), kairous, and temple halls for years now aiming for extreme realism. I already nailed the visual realism but I always encounter tiny details that causes the occasional revision. One recent one was the ronji-daruki. Although the ronji-daruki is not "tiny" in the sense of precise Japanese carpentry work, it is tiny in terms of visuals when seeing the building as a whole from a distant.

    As for the ronji-daruki, my "old models" always have both base rafters and flying rafters aligned. However, despite this, I figured that just widening the hip rafters was quite a primitive method. For the past three days, I made a lot of 3D sketches and observed how curves behave especially when nearing the eave edge and figured a new method to align the ronji-daruki based upon the projection angle of the kioi, the area along the kioi that receives the flying rafters, and the side of the hip rafter that intersects the kioi. As of now, I am confident that I am out of the dog days of problem solving as I have a completed revised 3D sanmon roof in place that addresses all the concerns mentioned within these replies.

    Interestingly, the new roof mentioned just now is part of a fourth sanmon I have done and the seventh Japanese roof I did altogether ("design and redesign" as was mentioned in the first paragraph of this blog which I agree in its entirety). Although I do concede that internet searches makes the learning arduous compared with reading books, I have enjoyed it very much.



All comments are moderated, so if you're planning to spam this, know now that your clicking and pasting is in vain. I do read the comments before posting, so your mission is doomed from the outset. All this time and effort trying to put your inane spam onto blogs -- is this how you want to spend your time on earth?

Please do me the courtesy of appending your name to your comment, even if posting under the 'anonymous' option. No name = deleted.

Comments NOT accepted include:

-those containing links unrelated to blog content
-spam of any kind, or ham for that matter
-did I mention that attempted spam postings will be non-starters?