Ming Inspiration (6)

Today marks the sixth post in this series describing the design and construction of a dining table drawing inspiration from a unique Chinese Ming side table. Previous posts are archived to the right side of the page.

After looking at Chinese tables generally in the first post, and corner leg forms more particularly in the second, and then delving into some of the framing and joinery options for the connection between the leg and upper table structure, we cruised in to the last post, where I went over the process of analysis that led me to conclude that bubinga would be the best choice for this project. Happily, the client also likes that material, and trusts the reasons for my choice (not all of which were explained to him at the time in interests of brevity). We have sourced the material with a supplier in Pennsylvania and it should be on its way to me in the next week or so.

Today I'll like to share with you some of the details of the design I have worked out for this table. Hopefully it will be clear to the reader where I have drawn inspiration from the table featured in post 4, and where I have chosen to diverge. Time will tell whether my choices have been good ones, though I feel confident in my route so far.

As mentioned in the previous post, a dining table brings with it certain aspects which will dictate some design differences from the Ming side table original. One of those aspects, mentioned in the previous post, is that the width of the dining table pretty much precludes the use of a single piece of table top wood in this design. I have accordingly designed it around two pieces of wood, both of which should be around 90% VG material.

Another factor here is the size of the table. At first the client was talking about seating for 10 diners, however a table of a size to accommodate that many people was actually too large for his dining room. We have since settled upon seating for 8. Normal practice is to allow 24" width per diner, and add an extra 12" at each end of the long side runs so as to allow room for the diner and his setting at the table ends. So, for 8 diners, we have 3 on each of the long sides and one on each end. Three diners on the long side, plus a foot of space at either end gives an 8' length. Measurements taken in the dining room of the client's house suggested it would be good to trim that length a tiny bit, and we have settled on a length of 7'-10", or 94", as I prefer to describe it. I always tend to use whole inches instead of feet and inches, even on large projects, as it eliminates, for me at least, a possible source of confusion and number inversion, etc, which can happen with 'imperial' measurements sometimes.

The width of this table is to be 40". With a table measuring 94" x 40", and the legs placed at the corners, we are close to the practical span limit for a conventional corner leg table, in terms of avoiding problems with the table sagging in the middle from its own weight and in terms of adequately supporting the table top while giving room for a diner's legs to fit underneath.

However, this is not a standard type of table with a solid plank top, though it may look like that. The design I will be employing, borrowed closely from the Ming example, slices abut 60% or so of the weight from the table, in relation to a top that was to be constructed from a thick slab. I've come up with another trick to reduce weight - more on that later.

I first showed the client a table design at the furniture show which was very much a preliminary sketch - here's how it looked:


Another view:


This table was of the un-waisted variety and was meant to be a drawing table or computer desk. Again, very, very preliminary.

After I started working with the client and in light of the dimensions of the table required, I sent him a 'mark-II' drawing. Now the table is about the right width, height, and length, and now has a waist, albeit a very narrow one:


At this point I was following the Ming side table in terms of having a support frame and a separate table top frame, components which would be attached to one another using stub tenons atop the legs and floating tenons along the mid-points of the length.

Further reflection upon the matter however led me to veer away from this approach. I had access to adequately large stock so that I could construct the entire side of the table from one piece of wood. This would be much stronger than a two-piece design. Given that the height of a dining table needs to be in the 29"~31" height zone, and that 24" of vertical clearance is required for a diner's legs, I am left with about 6" vertical measure at most with which to construct the side structural arrangement of the table.

In considering the primary long support rails for this table, of which there are three, one way to lighten them while retaining their strength came to me as I considered them in purely structural terms. A beam made of steel, for example, needs to only have material where it is needed -economy in material in other words. A common form is the 'I'-beam, in which you have steel oriented on the top and bottom surfaces so as to resist compression and tension, connected by a slender web of material. A wood beam behaves essentially in the same way as that 'I'-beam, and as past posts on that topic have indicated, so long as the top and bottom surfaces of the beam are left relatively intact, material may be removed from the middle, i.e., the neutral axis, and a large percentage of the beam's strength will be retained. This is the logic of the tusk tenoned (and here too) connection, for example.

Considering this piece, then, I could see that making the waist wider, while leaving the overall height of the outer long rail sections the same, would be akin to moving in the direction of the ideal 'I'-beam configuration. So, the frame now has a taller waist:


There was a balance point to be struck there in terms of deciding about the waist, given the vertical constraint of 6" for the long rails. The taller the waist becomes, the smaller the apparent thickness of 'table top' rail and 'apron rail' become. That balance point for me was an apron of about 0.5" in height, slightly above center, between the apparent 'top' and the apparent 'apron'. The 'apron' is left appearing to be a slightly thicker piece than the apparent table edge above it.

Another difference to the Ming side table with this piece is the treatment at the narrow ends of the table. In that side table, everted flanges were incorporated into the end pieces. Such flanges, while beautiful and serving to restrain a cylindrical scroll from rolling off the table top, are not appropriate at all for a dining table top. Still, I needed to cap the table top panel end grain portions, and provide some restraint against any tendency which the top may have to warp. For the most part, restricting the top from warping is handled by the vertical grain orientation of the panels, along with the transverse battens which are affixed to the underside of the panels with long sliding dovetails. I just needed some clean way of terminating the table panels at the ends- I settled upon a variation of the double mitered breadboard end:


The next significant point about this table in relation to the Ming side table is that even with the table top weight reduction and configuring of the long pieces to slice more weight, the fact remained that this table overall had a heavier mass than the inspiration piece did. So, I felt I needed to add some form of additional support, and as mentioned in an earlier post in this thread, my choices devolve to:

1. humpbacked or straight stretchers
2. banwancheng, or giant's arm braces
3. spandrels
4. some combination of the above

The most discrete option are the giant's arm braces. The typical ones seen on Chinese tables, stools, and so forth are serpentine, 'S'-form pieces. They are arranged this way on account of the method by which they are attached to the leg and the underside of the table:


The 'S'-form allows the grain of the brace to be more or less perpendicular to the leg where it attaches, so as to keep good grain orientation. The joint at the leg is formed by a sliding dovetail, and fixed in place by a plug inserted afterwards from below. On the underside of the table, the brace is pinned in place.

On square tables, the banwancheng can all meet at the center, and a form of 'cap and trap' is used (from Ecke's book):


I am trying to avoid the use of pegs if at all possible on this piece, and I thought that the 'S'-form braces, while having a pleasing form, had the defect in this case of being likely to interfere with diner's legs. So, I have come up with a giant's arm brace which arches upward, and drawn upon my background in Japanese carpentry to employ hiyodori-sen, or crossing and piercing wedge-pins, to attach the braces to the underside of the table:


The use of hiyodori sen is typically associated to attaching hanging jack rafters to a hip rafter, or reinforcing the connection of perimeter fascia where they meet atop the hip rafter. Here they provide a means to directly attach the giant's arm braces to the table's center beam, thus providing a very direct load transmission path. It's a strong connection linking end grain bearing surfaces, and no pegs in sight. The connection between the lower end of the giant's arm brace and the leg is accomplished with a fairly novel connection, which I will be detailing later on in the build.

If you examine the foregoing picture a little bit, you will note that the perimeter frame is thickened at the ends and thinner in the middle. Here my design follows natural forms - just as bones are thickened at their ends to provide more material at the connection points - the frame rails do the same. Another parallel could be drawn to the use of 'gunstocked' posts in English (and American Colonial) timber framing. This enlargement of the ends (or reduction of the middle, in actuality) of course allows the corner joints to be beefy without a concomitant weight penalty in the rest of the table.

After sending my rendition of the arched banwancheng to the client, I was most pleased when he got back to me with an enthusiastically positive approval. That's the great thing about collaborating on the design with a client who is willing to be involved and engage in the process.

Subsequent conversations with the client led me to explore further options for reinforcing the corner connections - like these spandrels:


He didn't like those so much, citing their obstruction of the view of the giant's arm braces, and neither did I care for them, for a variety of reasons. Mainly I thought them unnecessary.

Next I explored humpback stretchers, however I didn't like that direction much either. Then I tried a simple curved brace pair in each corner, used in concert with the giant's arm braces:


It's starting to look a bit like a tree here, and much as I do like trees, it wasn't a direction I was at all interested in, however I wanted to explore it on the client's behalf. Here's another view:


A significant aspect of the problem with these was incorporating them in terms of the assembly sequence of the top, giant's arm brace, and leg. Also, knee braces are problematic to join to other pieces in any case. In timber framing, pegs are often used to attach them, but the reality is that the peg's greatest asset is one of holding the braces temporarily in place for assembly of other components. If, in service, the braces are actually loaded by shear so as to create any tension at the joints, the minimal relish in the tenon leads to near-immediate failure at the connection. A lot of times in fact the driving in of a draw-bored peg will shear the relish (shh!). Braces work in compression, at least when the connections are all wood. Besides, I was really wanting to avoid pegging.

I was delighted when the client got back to me and said he didn't like the added braces, and that he much preferred the clean lines with only the discrete giant's arm braces. I was most pleased, if not feeling a bit vindicated, as that had been my conclusion all along.

So here then is an overview of the table after design is more or less complete:


Getting the leg shape sorted out was a most onerous task. I noted wryly to myself at one point, in the middle of some hair-pulling, that Nakashima never seemed to have had much concern for the legs in any of his slab tables - all you see are simple little turned spindles. I'm jealous - how easy could that have been? In all honesty I find the leg treatment on his tables to be one of the design shortcomings of those pieces - they look like legs sourced at Woodcraft or something like that. Anyway, please excuse that brief aside.

I found the legs quite a challenge to get right, and tried many variations - here's but three:


Finding the right point between ornate/organic and clean/modern lines was tricky, not for the least of which because it took a while to tease out from the client exactly what sort of look he and his wife were after. In the end I found something all of us liked. The legs and apron will further be integrated to one another by the use of a beading treatment along the lower/inside arris. Just drawing that bead in SketchUp was a royal pain, but I was able to produce something adequate for the purpose in the drawing. I explained to the client that certain forms were extremely time-consuming to render in SketchUp and that I would prefer to simplify them in the drawing. In lieu, I sent him photos of Chinese pieces with similar detailing so he would have an idea of what I was intending. I still haven't settled on the exact size and configuration of that bead at this time. Some things I leave for a point down the line in the making process where the physical appearance will be more defined - more so than a computer drawing - and I may draw fresh inspiration from that. Just having more time to mull it over often helps as well. I might even do a few mock ups, experimenting with different edge beading treatments.

Here's another picture showing more clearly the 'spine and ribcage' like arrangement of structural members on the underside of the table:


Another view of the table corner:


The banwancheng are only at the roughed-out stage in the drawing still - I will be refining their profile later in the actual build. Sharp-eyed readers may have noted certain things going on around the corner joints of this table, and I will reveal my joinery solutions for those corners soon enough. Also, there are joinery details for the transverse table supports that not even the client is privy to at this point. All will be revealed soon enough. This post is long enough already!

Thanks for coming by today, and comments are most welcome as always. On to post 7