First Light XXXVIII

Here at post 38, I’m thinking that this thread should finish off in a post count somewhere in the early 40’s -if it doesn’t finish me off before then! We are in the final throes of the job now, and this is a stage, for me, that is characterized by a battle against impatience. The project has gone one for a while now, as regular reader of this blog will I’m sure agree, and now that I’m down to the nitty gritty of a few grill bars, there is very much the temptation to just get ‘er dun Jimmy!

Well, that temptation to give into impatience is something I have learned to guard against from past experiences, though it remains a struggle for me at times. In fact, I know if I rush something now, and screw it up, necessitating a re-make, my patience will be tried even further. So, I actually try to slow down half a pace at this stage, and strive to remain focused and work in a methodical manner. These little, seemingly inconsequential near-to-the-finish line steps of putting the kumiko assemblies together with their frames in fact has been one of the most vexing steps in the entire project.

It may not be altogether clear from the descriptions of the design and build process of these grille assemblies for the lantern housing, just what factors in particular make them so complicated to put together, so I’ll try to explain. The frames are one aspect – given that they are splayed outward on the verticals, with a top and bottom frame rotated so as to be in plane with the floor and ceiling of the lantern housing. Then there’s the kumiko assemblies, which are a bit tricky to connect to each other with their mitered housed half laps, given that all the bars are on slope and the central pieces are thicker than the ones that surround them. Those two bits were sorta tricky, but not too bad really- it is when you try to put them together that the headache begins. The kumiko that slopes downward from the side frame to the crossover laps creates one set of angles at the join to the frame; the kumiko that slopes upward from the frame creates another. The top and bottom connections between kumiko and the frame are the easiest, being the some pair of angles (which are different than the angles for the side connections. The corners, where the central main ‘X’ bars meets, have a sword-tip miter with different angles on each side. All in all, you end up with eight different angles to deal with, some of which are only a few tenths of a degree off of 90˚.

I now realize why the vast majority of there lanterns – I have looked at around 90 examples – almost invariably feature kumiko in a regular rectilinear grid pattern, three up and three down. Such a pattern would be so much easier to fit into a splayed opening. Many of these lanterns in fact do not have a splayed post type of housing, which makes the kumiko sets no more complex than regular shōji.

Yet that three up-three down pattern of course offers little in the way of shear load resistance, and the non-splayed openings don’t look nearly so nice as the splayed ones and mean that the roof atop a non-splayed lantern housing either must be undersize or have disproportionately large overhangs, or the lower portions of the lantern have to be made overly chunky to provide a larger housing. I’ve seen all of these outcomes in the Japanese lanterns I’ve examined. None of those apparently simpler directions were of interest to me, either aesthetically-speaking or from a structural-logic perspective. There’s no doubt in my mind that a splayed lantern housing is miles stiffer than a non-splayed one, and it better carries the roof in a graceful manner. There’s also no doubt in my mind that using the diagonal kumiko arrangement with the type of surrounding framing system in this lantern housing was the best choice too – next time though I might consider a different kind of structural framing arrangement that would not require the additional bracing function from these grille assemblies.

I’d made my choices in this lantern project for sensible enough reasons, I feel, however it is also a case of now having to lay in the bed I’d made, so to speak. It was good to come to a point of grasping the implications of my design decisions a bit more clearly – this was really only possible after reaching this stage of the construction process. I hope this provides the reader with a good illustration of one of the downsides of dividing designing and building from one another. Only by building my design was I able to come to a deeper knowledge of the pros and cons of the design from a constructional standpoint, and understand better the forms I have observed as common in Japanese lanterns. If I merely designed, and someone else built, numerous small points of difficulty in the construction would likely be lost to me, and the builder would likely not understand the reasons for some of my design decisions. Design AND build form a self-reinforcing loop, which inevitably results in better outcomes – that’s my feeling on the matter at least.

If I make another lantern, I can bring these lessons forward and improve both the design, and streamline the build.

I was able to figure out what the angles of the various kumiko-frame connections were, however that was really only a peripheral problem. More to the, er, point, was how to connect the kumiko to the frames- given that I wanted a good cosmetic fit and wanted a mechanically strong connection. The whole idea of employing diagonal kumiko, with the central set forming an ‘X’ brace, was to provide good torsional shear resistance for the lantern housings. They could not simply be wedged into position by friction fit, as is often done with certain sub-patterns within complex kumiko, such as are seen on high class shōji and ranma.

The grille bars had to mechanically locate and have good bearing surface and give a clean interface where the faces are easily viewed- the compromise between these three points seemed to be in a bare half-lap at the end of each bar. Further, for a clean joint interface that would disguise seasonal wood movement, and a chamfer on this inside corner of the frames, I now chose to recess the kumiko back from the front of the frame pieces by about 1/16″ (about 1 mm).

Given the slight variances between frame member thickness, position and rotation of the housing posts, etc, each grille assembly was slightly different from the next one. I couldn’t readily scribe the joint intersections, given the frame having top and bottom rotated one way, the sides another, and the fact that the kumiko were to be set back that 1/16″. I knew that any error in laying out the location a mortise would look bad and a poor abutment would also mean the joint wouldn’t perform mechanically as it was intended. I was getting that ‘painted into a corner’ feeling.

After much head scratching and hesitation, I realized I could use the fixture which I made the kumiko assemblies to also cut the half-lap tenons along the sides. The side frames, though sloped, were parallel to one another along their inside faces due to the fact that I had backed the inside of the posts earlier in this project. so, I placed the kumiko assemblies back in the fixing jig and used a pattern routing bit with an MDF scrap as a guide and trimmed the sides of the kumiko in place. That still left the top and bottom cuts to figure out however, and I definitely didn’t want to cut them too short as that would ruin the kumiko altogether. I wasn’t feeling much like sawing-cutting-planing-squaring and routing another set of kumiko, despite how much fun they had been to make. I knew I needed to tread carefully.

So, I paused and mulled things over a little more. Then I thought, “well, at least I can mill the grooves into the frames to accept the glass”. I don’t have the glass yet, so I called up a local glass place and asked them if they had any thin frosted glass (‘yes’), and how thick it was – the customer service woman said “0ne-eighth of an inch”. I tried to ascertain how precise that number was, however she didn’t have that information. My paranoia in this regard comes from prior experiences in Canada, where glass is often sold as having certain nominal imperial inch measurements yet is actually made conforming to metric standards. I had no idea whether this practice was confined to Canada or not, but the last thing I wanted to do was to put together a frame – and yes I must glue this frame up – so that the groove was too tight to accept the glass. Making a sloppy loose groove that left the glass rattling in the wind was hardly desirable either, so I grabbed my calipers and drove down to the glass place to measure for myself, just for piece of mind.

I was pleasantly surprised to find the glass measured exactly 0.1250″ – damn! that’s some accurately made material! I like that, and now I’m curious to learn how the glass industry can produce a material like glass – a liquid no less – to such high tolerances.

So, back home I went, inspired by the accuracy of the glass industry, then set up my router table with a 0.125″ slot cutter to mill the grooves for the glass. The glass by the way will cost about $8/piece, which is pretty reasonable I thought. I then proceeded to mill the grooves in the one frame I was working with. Two passes, raising the cutter slightly between runs, gave me a groove that was about 0.02″ fat from dimension, which should allow for a easy enough slide-in of the glass panel and hopefully no rattle.

I then went back to my sawhorse/workbench and started puzzling out how I might best make the cuts on the top and bottom connected pieces of kumiko. Then I noticed something odd about the grooves, they seemed to be in the wrong place. I soon realized my blunder – without realizing it, I had had the frame sitting on the work surface upside down – the entire morning I mean! Thus I had made the grooves referenced from what I thought was the front face which in fact was the back face of the frames. Oh dear. The grooves were thus off-centered the wrong way, and the frame was ruined. Or so I thought.

After a cup of tea and a little break, I was able to go back to look at the situation in a more calm state of mind. I figured I’d have to make another frame, which I was definitely feeling a little annoyed about. Then an idea came to me – while the groove was in the wrong position for the glass, I could slightly widen it so as to allow the kumiko half-tenons to fit. Then I would have a means of zeroing in on the correct cut placements for the kumiko half laps. Since the frame was presumably ruined anyhow, it certainly wouldn’t hurt to explore the situation a little further. I returned to the router table, and cranked the cutter up a bit an widen the slot. Then I made some half-laps, staying wide of the line, on the kumiko, and did some trial assemblies, working the fit until it was satisfactory between the grill and the frame:

Here’s a view of another corner:

It then occurred to me that the ‘mistake’ of putting the groves on the wrong side of the frames had actually resulted in the solution to this tricky fitting problem: I could simply patch the portions of the grooves with infill strips, leaving ‘mortises’ for the half-lap tenons. This way I could obtain very accurate fits at the joints, giving the structural performance i was after, and only suffer marginally on the aesthetic side, as the infill strips should be pretty much invisible if done cleanly. Further, when I mill the groove for the glass (uh, in the correct place this time!), the rear glue line of the infill strips will disappear altogether. Whew! Problem solved.

This approach to building the frames up is not a particularly fast one, as it takes me most of a day, believe it or not, to fit up a single panel of fame and kumiko. So far I’ve got a couple done – take a look:

Another view:

These parts haven’t received their final planing or chamfering yet, so I anticipate it will finish out a little cleaner than at present, but all in all I’m satisfied with the result. As for the glue, well, yes, it was my intention to avoid glued construction in this piece, however given the small dimensions of the involved components, and the need for rigidity in the assembled panels, along with the arrived-at solution of inserting patching strips, I feel I must use glue for these pieces. So I fall short of my goal in that regard, but I can definitely live with it. If the glue fails down the line, the panels are fairly tightly trapped, dadoed within a fully joined, glue-free structure of posts, plate and sills, so the use of glue here is decidedly non-critical to the integrity of the lantern housing as a whole.

Here’s a few more shots of the lantern as it stands this morning:

From the side:

Just two more panels to put together – one of them is the same as the other two already completed, while the last panel will be a little different, as it is designed to be easily removable (so the light bulb can be changed) – again using only joinery. That will be the subject of the next post in this thread.

One last picture of the two installed grille panels:

Thanks for your interest in this blog- I hope to see you next time, post 39.