Sage-Kama Desk – Build 3

We left off the desk build last week with the segmented dovetail joinery on the aprons/cross pieces and underside of the desktop. One thing I failed to mention last time was that before cutting the okuri-yose-ari in the desktop, I finished planed that surface. Planing after cutting dovetails means you run the risk of loosening the fit of the dovetails, since you’re essentially moving the shoulder of the joint with each pass of the plane, making the joint looser. So in general, hand plane before cutting dovetails!

The next bit of joinery to cut was the sage-kama joints (one sided dovetails) on the ends of the leg stretchers and the wide cross pieces. I also cut the round cutouts on the ends of the wide cross pieces which you can see below. I used a tablesaw jig to make the angled cut on the end of these parts, and after knocking out the waste I used a chisel to clean things up. (back in the first post in this build series I cut the mortises for these joints in the legs).

The sage-kama joints were the last bit of joinery to cut, so once those were done I moved on to handplaning all the parts, starting with the upper side of the desktop.

Here’s the plane I used this time, it’s an Ishido 60mm Swedish steel blade. Hollow ground and double beveled. Some folks might cringe at that, but I’ve yet to find a more consistent and quick way to achieve a sharp edge. On hard woods like this white oak I really like using a slightly narrower blade like this 60mm. Standard kanna at 70mm wide are a lot of work to pull, and when you have a planing session that lasts a couple hours or more, it’s nice to use something that doesn’t kill your stamina.

After planing the top, I pulled out my mentori-ganna and chamfered the edges. You can see the round cutout for one of the legs as well, where it will be inset into the desktop. At this point the cutout was roughly shaped, and I later fine tuned the fit after dry fitting the legs.

Here’s my mentori-ganna, a Yamamoto made plane.

I also planed all of the stretchers, aprons and cross pieces, again using the same 60mm Ishido kanna.

With all the parts planed, I moved on to assembling the understructure starting with joining the long aprons and intermediate cross pieces. The beautiful thing about using mechanical joinery like the wedged tenons on these parts, is that there’s no need for glue, and the associated stress that comes from having to get things clamped and tightened before the glue starts to set.

Once the shoulders of the joints were tight and with the joint still clamped down, I tapped in the wedges. I did use a touch of hide glue on the wedges to prevent them from coming loose. Technically these joints are reversible once you remove the wedges, and I’ve had the opportunity to do just that when working on some old shoji screens. Hide glue makes the joint a fair bit more serviceable than glues like Titebond, since the glue bond can be more easily broken. That said, I’m guessing these joints will never need to be undone (fingers crossed), but it’s nice to build furniture with that potential in mind.

Here I’m clamping the wide cross pieces onto the ends of the long aprons.

Followed by some more wedges.

Here’s a view from the opposite side, with all wedges installed and the understructure fully assembled.

Once the glue dried I trimmed the tenons using my router. On the wide cross pieces I left the tenons proud by about 3 mm.

On the intermediate cross pieces, I trimmed the tenons flush. When I was milling the stock for all these parts at the beginning of this project, I made a decision on the spot to leave the long aprons slightly thicker (~2mm) than what I had planned, but I failed to take that into account when I cut the length of the tenons on the intermediate cross pieces before assembly, and so truthfully I didn’t have enough material to leave a matching 3mm reveal on the tenons like I did on the wide aprons. So I cut these flush. I actually like the result thought, with one set of joints cut flush and the other proud. It’s an interesting contrast and I figure it’s also an opportunity to show customers both approaches.

With the understructure assembled and the top planed it was time to put them together. This was the part that I was worried about back when I started designing this piece. Bringing 4 sliding dovetail joints together at the same time, is not exactly the most straight forward task. Using the segmented dovetails definitely made things a bit less cumbersome since each cross piece only needed to slide about 30mm, but I still had to work carefully moving each piece little by little. The real problem I ran into though, was simply that the joints were too tight. I intentionally made the dovetails on the tight side, with the goal of having a slight compression fit, but I could have definitely shaved a bit more material off because clamping and tapping these parts into place was a bear. In the previous post I showed a test piece that I used to gauge the fit of the dovetails, and while the test piece was able to go in relatively easy, I think the fact that it was just a portion of the joint, and not the full length, meant that I wasn’t gauging the full amount of friction that I would be fighting against during assembly.

I started by using a series of bar clamps and cauls to press/slide the cross pieces into place, but ultimately what work best was using a dead blow mallet. All in all I was clamping and pounding on the four cross pieces for at least an hour trying to bring these joints home. I had more than a few moments where I started to wonder if I was fighting a loosing battle, and would need to take things apart to loosen up the dovetails a bit. But in the end I got the joints closed, with no damage done to any of the parts.

It’s a pretty rock solid fit. There was a slight curvature to the desktop both along and across the grain, and both flattened out completely once the dovetails were engaged.

One of the final steps before finishing was to install the legs to do a bit of final tweaking and fitting, mainly in regards to the half-round cutouts where the legs inset into the desktop. Earlier I rough cut those areas, and with the desk now assembled, I was able to do a bit of scribing, making each cutout match the shape of it’s respective leg with a slight open space all the way around.

Here’s how that goof I made on the first set of segmented sliding dovetails turned out. I overshot my reference line when I cut the notches (see previous post), so you can see a hint of those gaps along the edge of the cross piece. It’s kind of a bummer but in the end it’s pretty inconspicuous and doesn’t effect the strength of the joinery.

With all the parts fit, trimmed and adjusted, I disassembled the legs and stretchers, and started finishing the wood. I contemplated leaving the wood natural, but we already have a few white oak pieces in our house that are natural and I wanted to try something different. I’ve always loved the look of ammonia fumed oak, but with seriously nasty fumes involved, it’s not exactly something I can do in my current shop space. So I opted for a more traditional approach using kakishibu and washing soda. Kakishibu is a traditional dye here in Japan made from fermented persimmon juice. Applying kakishibu saturates the surface of the wood with tannins. The color from kakishibu alone is kind of a dull pink-ish tone, and not super attractive. So once the kakishibu dried, I wiped on a washing soda/water mix which reacts with the tannins and turns the surface a rich brown color. I really love this color on oak. It’s certainly possible to achieve a similar result with an off the shelf stain, but there’s something really pleasing about using a more traditional stain made of simple materials. Once the surface dried completely I applied a finish coat of oil to all surfaces of the desk.

After that I packed up the desk, brought it home, assembled it, and used it for about a week. But I quickly realized that having one stretcher along the long length of the desk, was not enough. The overall structure was stable, but it was really easy to rock the desk side to side, with just that one stretcher.

Below is my original design as I constructed it, and you can see the single long stretcher joining the back legs. Front to back the desk is rock solid since the legs engage with both the wide cross pieces as well as the short stretchers. But that single stretcher running along the back didn’t provide enough racking resistance.

So after playing around with a few different ideas, I decided to add one more stretcher above the existing one, as well as a panel sandwiched between the two. The panel would essentially make the two stretchers a solid frame and add a lot of racking resistance. The design on the panel below was something I quickly whipped out for the sake of visualizing the concept.

Next week I’ll share the details on the modification of the back legs and the addition of the extra stretcher and panel, and with that this desk build should be complete. So stay tuned and thanks for reading.