Wednesday, November 6, 2013

Gateway (III)

Much has happened since the last post in this series, from some three weeks back I guess. I went to Oregon and looked at a bunch of Port Orford Cedar logs. I have met the gardeners involved in the project and now find myself in a bit of a war. Not something I am going to get into discussing here, and what has devolved from this situation is that the design of the new gate remains an open question. As in 'unsettled'. In fact, 'open' in more than one way too as the Museum of Fine Art in Boston now says they are giving freer rein to the design. This is good, despite the lingering uncertainties otherwise, as the previous design I showed (last post) was as much a compromise as anything else, trying to fit between the various competing visions and interests in regards to the new gate. Now that some of those visions need no longer be satisfied, I can take the design in a more 'pure' direction. While the previous iteration was a low pitch mukuri (convex) roof atop a framing system that was a cross between that seen on kabukimon and yakuimon gate patterns, I can now move into a gate which is really purely a yakuimon, the 'physician's' gate. And the mukuri roof, while a form which is likely my favorite among the many shapes and styles of Japanese roof, turns out, after some further research, to not be entirely suitable in this setting. So, now instead of a convex roof body, I am designing for a concave roof body.

There are several words in Japanese used to describe the shape of a rafter, and the associated roof plane which follows it, when made concave:

照り(teri): the usual expression for concavity, and one which would be paired with mukuri when describing concave/convex roof forms in general, as "teri-mukuri ya-ne".

弛み (tarumi): this indicates sag, which is one way to look at making a rafter concave - pull the middle of the stick down relative to the ends.

反り (sori): this indicates upsweep, which is another way of looking at making a rafter concave - pull both ends up relative to the middle of the stick. Sori is more often used to describe curved hip rafters, which are almost always, in Japanese roof framing, curled up from the line of the theoretical straight hip.

I adore Japanese roofs in general. They are the important architectural element in most Japanese traditional wooden buildings, most especially temples and shrines. As I've mentioned before on this blog, the word in Japanese for roof is ya-ne (屋根), which literally means the root (根) of the house (屋). Without a roof, you don't have a dwelling.

For the many years I have been studying and, occasionally, building Japanese roofs, my fascination has been primarily with hip-roofed forms, and learning the geometrical methods to produce these forms has been of keen interest. I was less interested in gable roofs (kiri-zuma yane) generally, and tended to think of them as I suspect many carpenter might, as essentially being fairly simple. I prefer the visual of a continuous eave line right around a building, and consider hip roofs to be inherently stronger, so that is where I have focused my attention for quite some time now.

Well, 'simple' as such is not really the case at all when it comes to Japanese gable roofs, except insofar as their appearance - and this is particularly true if they have curvature in the body of the roof. I have found the 'simple' and clean appearance of such roofs conceals an underlying degree of complexity that is quite astonishing. I shall underestimate these no more.

The curves, you see, make the curved gable roof very much akin to a boat hull, and just like boat hull construction, there is lofting and layout to get you so far, and then there is fairing the curves to reach the desired outcome.

I have a predisposition towards finding a clean geometrical means to lay out and construct a roof, not interested in the cut-and-try, fudge-and-fit methods so often seen. I hold to the idea of all the components of a roof being laid out at ground level, cut to the line, and when positioned in the roof assembly, clicking into place like a German car door, without question fitting perfectly. That ideal and reality, however, don't always meet so tidily; nonetheless that is the possibility to which I strive toward.

With the curved gable, while geometrical layout techniques will get you most of the way there, what you will find is that there are intersections between parts that cannot be cleanly resolved on a drawing. There are areas where surfaces must be 'faired' to one another judiciously. It took me quite a long time to realize this, and much hair pulling in the interim as I tried to get curved parts going in different directions and with different slopes to mesh together on the drawing. I'm surprised I have much hair left at this point frankly.

I'll show you a simple example first - here is a view of the corner of a gabled roof build up, the upper level called fukiji - look closely at the corner where the boards meet:

It looks at first glance like a simple goof up in cutting, however I have found that when using only 2D drawing it is quite difficult to eliminate these little offsets. And note that the above offset is among the most minor tricky bits at this corner of a roof.

Directionally, geometrically, there's a lot going on at the corner:

It's a really interesting procedure to delve into a roof design when a certain amount of it is flying blind. I have most every Japanese carpentry layout book in print, and a good number of shrine and temple architecture books, pattern books and so forth, and it is curious how little information one finds on kirizuma roofs. You will find the odd sketch or two showing the proportions of the barge board on a gable, and some of the geometrical techniques to produce their curved shapes, however the gables in question are invariably part of hipped gable roofs, and are not pure kirizuma.

This might seem like a small matter, however there is a key difference between a hipped gable roof's gable, and a pure gable: the hipped gable's barge board's lower end terminates on the body of the roof and there is no clear visual tie to the rafter pattern, while the gable roof's barge board finishes at the edge of the eave, and the end of the barge board is part of the pattern of rafter tips visible along that eave.

This still may seem of little consequence until one realizes that the gable barge board, or hafū ita, on a Japanese timber building is not plumb. The gable's barge boards are in fact tilted forward - termed korobi - as part of a sophisticated set of aesthetic tunings which reduce the visual foreshortening effect when looking upward at a surface. To be clear, the barge boards do not project forward at the peak of the roof like a prow, rather the entire unit of both barge boards is tilted forward. This might sound jolly nice until one starts to look at the connection points between the barge boards and the rest of the frame. At every purlin, you have compound abutments, and at the eave ends, where perimeter fascia stack up, and may also curve up and if so also usually swell in thickness, you have a bit of a nightmare where the parts come together.

But wait, there's more! One sophistication of Japanese roof work is the creation of visual effect, such as having very light appearing rafters which are in fact a soffit. The true rafters are within the roof and usually unseen. Another trick is to create the appearance of a thick roof edge. Thin roof edges connote an insubstantial roof, and are generally only seen on cheaper work and some modernistic stuff. We won't go there. A thicker roof looks much beefier, however it is also desirable not to burden the roof structure with excess dead loading. So, there are methods of building up a thick roof edge, an application generally termed noki-tsuke.

One way to build up an eave edge is with a stack of wood shingles, which are quite readily chopped and planed into the desired form along the front face of eave build up thus attained. However with a copper shingled roof the wood-shingled build up is not appropriate, and instead a tilted board on a support piece is added (refer to the above picture). The tilted board will eventually be shingled in copper.

This tilted board is called the fuki-ji, and as it has a different front face tilt than the parts immediately below it, there is a curious effect which occurs as the eave curves upward toward the corner: the fukiji board wants to deviate outward from the support piece. The dreaded 'fukiji problem'. I wrote about this issue a few years back when designing the Japanese bell tower (←link) and I had forgotten about it, but sure enough it reared its pretty head. Nice to make your acquaintance again my friend.

Once it became apparent the same problem I had seen previously was presented afresh, I was able to solve it readily enough. Been there done that. Essentially there are four different solutions to this problem, one of which is a bad solution leaving only three choice really. I went with what seemed like the reasonable path: curved the fukiji upper board inward so that it follows the support board and maintains a parallel line to the piece below when looked at from below.

After the intial redraw, I produced this roof shape, viewed in elevation:

Note that the minoko are absent from the roof at this stage - they are very time consuming to draw and I wanted to look at the overall form before getting into penciling in all the details like that. The above rendering was not too bad, but right away I could see something had gone awry when it came to tilting the board boards outward -these had way too much forward tilt. Not sure what happened there, but it was unintentional. I'll blame SketchUp, as I'm used to doing so.

I needed to do a complete redraw of the entire gable end assembly, but in the short term I decided to reduce the tilt of the boards so i could see how things looked.

Here you can see the effect of that re-tilting backward as the connection points at the corner, throwing the planes slightly out:

A while later the overall shape of the roof was coming together nicely, at least to my eye:

I was planning to get back to the drawing the next day to complete the work, thinking that this was shaping up nicely. If I can draw it, I can build it.

Then I got an email from the MFA indicating that the decision had been handed down: the plan was to simply replace the existing gate with one that was the same design and appearance, with minor foundation and flashing improvements. Like that, it was bye-bye roofed gate. A mirage I guess.

Glad, I suppose, that I didn't put huge hours into the drawing - though I learned some very useful information in that exploration - and sad too that the decision makers were pressured away from doing a construction direction that did justice to the magnificent Port Orford Cedar 300~400 year old logs on hand. 'Political' considerations can trump what are otherwise rational decision making processes and a view to the 'long now', which is regrettable but this is how the world works folks. I'm glad to have the project, even if that project isn't what I consider to be the best course from a carpentry perspective. It's a rare opportunity to build a traditional Japanese structure, and that is all good as far as I'm concerned.

At the end of the day, I at least can say that I did my very utmost to steer them towards a roofed gate, and was nearly there. Close! So damn close....

So, the project is to build a kabukimon pattern gate after all. It will be fun project, and a much simpler piece than a roofed gate, and the time frame remains in place with installation slated for April 2015. That's plenty of time to complete the work; in fact, most of the time will be spent drying the wood.

I will of course update this thread as things move along. I've already drawn the kabukimon replacement, and expect only some minor tweaking as far as the flashing detailing goes at this point. Here's the gate I will build, more or less:

Thanks for visiting the Carpentry Way. Comments always welcome. On to post IV


  1. Chris;
    It is to bad you didn't get the redesign.Atleast you still get to build something Japanese! Good luck and god bless! Congradulation!

    1. JT,

      thanks. I appreciate your message, and you're right, at least I get to build something Japanese. It's a cool project.


  2. too bad about the roof!
    On an unrelated topic however, whilst browsing through a couple of furniture catalogs and such, i came across this and thought you might like to see it

    1. Damien,

      yes, I've visited that site before and the fellow does excellent work. I also had an idea a while back to do furniture pieces using techniques from the Mazerolle work, however I can see now this is being done very well by Voillet. Some clean joinery for sure.


  3. Chris

    One wonders how much poorer we would be, if Michaelangelo hadn't had the de Medicis.



All comments are moderated:

Anonymous comments will be deleted

Spam will be deleted

Comments containing links unrelated to blog content will be deleted