One of the furniture designs that I’ve been working on recently is this shelving system that I designed to be easily assembled/disassembled via traditional joinery. I see a few benefits to knock-down shelving. One, it would make moving easy by allowing the whole system to be broken down into simple parts that can be compactly packed and stored, versus having to physically move a huge bulky shelf that is permanently assembled. Two, it would make shipping easier and less expensive and it would allow me to flat pack and ship to customers.
My original concept consists of shelving as well as a variety of small cabinets that could be arranged according to a persons needs.
So the big question is what kind of joinery to use for knock-down shelving. My initial choice of joinery for this design was a joint that I first learned about when reading Chris Hall’s blog many years ago. In Japanese it’s called a watari-ago joint, 渡顎 (I’m not aware of an English name.) The simple animation below shows the parts of the joint and how it all fits together.
The name watari-ago refers specifically to the notch in the cross piece which mates with a matching cutout on the post/vertical piece. The joint is usually used to lock together two pieces that cross or overlap one another.
Working with Somakosha, we used a simplified version of this joint on almost all of the home building/carpentry projects that we did together. On traditional homes the main structure of posts and beams lacks any kind of diagonal cross bracing. Instead after the building is raised thin boards of wood, called nuki, are passed through the posts (through pre-cut holes) and are secured in place using wedges. The nuki span as many posts as possible and when they are wedged in place the entire structure really tightens up.
The thin boards that are passing through the posts in the image below are nuki from the Little Lights project. If you look close you can just make out the wedges on the top side of the nuki where they intersect with the post in the center of the image.
This type of joint is pretty similar to the joint in the animation above, minus the watari-ago part. Meaning there is no notch on the bottom of the nuki to lock directly into the post. The nuki pass through simple mortises and are simply locked with the wedges. Later bamboo lathe is attached to the nuki, followed by multiple layers of earthen mud/plaster, creating a really solid wall. That said there are certain construction tasks in which a nuki will be mechanically locked to the post using the watari-ago style joint.
So to bring this all back to shelving, I think there are some things that can be borrowed from this style of construction. One being that the nuki, or in the case of shelving, the shelf supports, can be simple boards with minimal joinery involved. With no tenon and shoulders to cut, the shelf supports would be easy to make and require minimal detail work. A simple notch cut into the bottom edge of the shelf support where it locks to the interior of the mortise is all that needs to be cut.
The mortises on the other hand are bit more involved to cut. The top side of the mortise also needs to be cut at an angle to account for the wedge, and the bottom needs to have a matching cut to fit the notch in the nuki/shelf support.
One aspect of this joint that I find slightly unappealing is the orientation of the wedge. The wedge needs to be driven in from a horizontal direction, and friction is the main force holding them in place. If you tap them in lightly with a hammer they do stay put. In the 6 years or so since I made a lamp using this very same joint, I’ve never had one of the wedges come loose, or worse fall out and get lost. But the possibility that the wood will expand/contract resulting in a wedge working itself loose, has always kind of bothered me, and it’s one of the reasons why I keep searching for another alternative to this joint.
So here’s one possible alternative. A hana-sen joint, 鼻栓仕口, or frequently called a tusk tenon joint in english.
This is probably quite familiar to many of you. I’ve seen this joint used on trestle tables and shelving for quite some time now. It’s a simple joint and solid. What I really like about it is that gravity keeps the wedge in place, and even if were to work it’s way loose, it’s not going to fall out. What I don’t like about it is that it’s much more involved to make. In the case of my shelf, the shelf supports would all need to have tenons cut on both ends. The tenons then also need to have an angled mortise to fit the wedge. That’s a lot of work. It also prevents me from being able to use shelf supports that span multiple vertical supports.
Lastly, for this joint to be truly strong, the tenon needs to be quite long to avoid any risk of the wedge blowing out the end of the tenon when being driven in (they probably should be longer than what I drew in the animation above). So you end up with extra long tenons sticking out from the sides. On a shelf with lots of joints, all those long tenons and wedges start to look pretty busy.
So what are some other alternatives? One more variation of the wedged nuki joint is this one, which we used on the engawa expansion project (片下げあり, kata-sage-ari, which translates roughly to “one-sided drop down dovetail”). With this joint the end of the nuki, or shelf support, has a half-dovetail cut into it, with a matching half-dovetail mortise in the post. The wedge then forces the male and female parts of the joint to lock together. The half-dovetail on the nuki aren’t that hard to cut, but the mortises definitely become more complex and involved. But on the plus side there are no protruding tenons making for a really clean appearance.
Compared to the previous two options, this joint is also much weaker against racking. There is simply much less contact between the post and nuki to resist any lateral forces that want to push against the structure. Nonetheless it’s still a cool bit of joinery, and could be another potential option.
I’ve been doing plenty of head scratching but so far haven’t come up with any other suitable joinery options. If any readers have any ideas, I’d love to hear them. For now though, I think I’m leaning towards using the original watari-ago joint. Cutout of all the joinery is the most straightforward and it’s definitely solid. I’m planning to show and sell some work at an upcoming event in Nagano this July and I think I’ll make a small version of this shelf for that show. I’ll be sure to share some shots from the build process so stay tuned.
Joinery aside, I’ve been looking at a few other shelf designs lately for inspiration and thought I’d mention a few that I really like. One cool design is this shelf from Karimoku, a furniture design firm here in Tokyo. I like this shelf a lot. It simple and clean, but still has a very structural feeling to it, and I believe its assembled with true mortise and tenon joinery. The open spaces where the shelves break in the middle for the two central posts to pass through, as well as the ends where you get a peek at the structure supporting the shelves give the whole piece a light airy feeling. All the components, the posts, shelf supports and shelves, are all really thin and slender which is a nice design, but it’s probably not capable of being completely loaded up with books.
Another fantastic shelf design, that I recently stumbled upon, is the 606 Universal Shelving System designed by Dieter Rams. The 606 system is pretty different from anything I’ve been visualizing lately, and as a system it’s really quite different from a stand alone piece of furniture. Nonetheless I’ve been looking at this design quite a bit and there are some really great features. I have no personal experience using this shelving, but judging from pictures/videos, the overall system is easy to assemble and very adaptable for different needs. Aside from shelving, it can also be expanded with cabinets and desk space as well, which is pretty cool.
Designing a system as elaborate as this is beyond my current scope, but studying this design has served as a good reminder about the importance of keeping things clean and minimal, while leaving room for the user to adapt the furniture to their own lifestyle. I think that’s really what this shelf system represents; very user oriented design. And that’s something that I’m trying to think much harder about lately.
One final design that I want to mention, and which has probably been the most influential is this shelf designed by a friend from my old community shop in Brooklyn (Studio SNNG). The shelf is designed using a similar system as the watari-ago joint, but instead of the shelf supports locking to the posts themselves, the short and long shelf supports joint to one another and are secured with wedges. It’s a pretty brilliant design, and truly breaks down into simple components.
Well that’s all for now. When I get started on actually diving into the making of this project I will post about it here. Thanks for reading.
It’s an interesting challenge to make KD furniture without metal fasteners yet keep a light and spare look. I can suggest a couple of other joints that might work. One is a through version of the kata-sage-ari. I have used it on stretchers for sawhorses. It is a little beefier than the blind version. It might be slightly easier to cut out than a watari-ago and a bit stronger since you don’t need much relish on the end.
Another might be a housed sliding dovetail. I think it is called an okuri-ari in Japanese. With a a couple of router jigs set up it isn’t that hard to make and is either completely hidden or can be made so with a small plug. No wedges needed.
The through version of the kata-sage-ari joint would be another good option and definitely stronger than the blind version and easier to cut. I’ll definitely keep that in mind as I move forward.
I considered a sliding dovetail, but I don’t really like having the exposed opening/cutout that exists after the dovetail is slid into place. And I could do a plug to cover those openings, as you mentioned, but the plug would have to be readily removeable to disassemble the shelf, and I’m not quite sure how to make that work. I think if I have to have additional parts, ie. wedges or covers, I’m more inclined towards wedges since they will be pretty simple to make and also serve more of a functional purpose.
If you cut your sliding dovetail male part on the just the lower half of the rail, the shoulder on the upper half of the rail covers the opening. No plug or extra cover needed. That does reduce the strength of the joint but maybe with multiple rails the sum would be strong enough to hold the shelving together.
On the other hand, excavating the female dovetail part on the uprights might remove too much wood if you want those uprights slender.
This one of those cases where I can see it in my mind but would need to cut a few joints at dimension to convince myself it would or wouldn’t work.
It would be interesting to do a test and see how strong this kind of joinery would be.
I’m leaning away from dovetails for a couple other reasons. One being that repeatedly installing and removing the dovetail will tend to compress the grain making for a loose fit. I think dovetails work great when they are installed once and left as is. But in my experience they tend to get looser and looser the more you assemble/disassemble them.
The other reason I still prefer the watari-ago/nuki style joint is that I think it will make assembly more straight forward. With dovetails the shelf stretchers all have to be installed on one end of the shelf/post assembly, and then that whole assembly has to be slightly lifted and the opposite end dovetails all have to be installed at the same time into the opposite end assembly. With the watari ago/nuki style joint the shelf supports can completely slide through the mortises, so you can install one shelf support, connecting it to one end and then the next to lock together the basic structure. From there you can slide in the remaining shelf supports one at a time.
Good discussion. Your thought about the dovetail not being suitable for repeated assembly is one I hadn’t considered but I agree. And I wasn’t thinking about the problem of assembling multiples.
Looking forward to how this design progresses!