After 2018, I’m working on a theory that odd-numbered years are better than even numbered years…..
I have decided, and not recently, as a result of numerous personal quirks and failings, that by and large I am not taking the simplest route, the quickest out, the fastest way from a to b or anything along those lines, particularly when it comes to woodworking . I’d rather realize an ideal that comes out of having taken a good long look at some of the best pieces from antiquity so that I can also ask: what’s the best answer to a given problem, given any constraints that might be present?
I’m not saying I have all the answers to that by any stretch, but am merely an observer and reporter of the scene along the particular lines of investigation that appeal to me. Hopefully some out there might also enjoy what might at best amount to rambling investigations of a joinery nature?
Besides the functional and constructional aspects relating to design, a piece built to last should look good – good enough at least for people to want to keep around for their lifetimes – and hopefully realize an outcome, in terms of ultimate durability, well beyond one or two lifetimes. Designing something to be robust is not so hard, but making a piece robust and yet graceful, not overly thickened in the part dimensions, well, that is an altogether different challenge. How to bottle the genie of ‘timeless design’ is another question.
With the futon cabinet, once I had worked with the client to determine the design direction and the rough sizing of the cabinet, there came the question of how the piece should look. Part of that is the choice of wood, part of that is the decision about how much joinery to express/hide, part of that is the nature of what associates to joined work. Finally, there are molded profiles and chamfers, which contribute greatly to how a piece looks, to play around with. Joined work is not going to look like veneered work most of the time, unless, that is, that the veneered work is making an extra effort in the artistic direction of resembling a joined piece of solid wood. This does happen.
If I am to, er, adhere to that concept of no-glue wood framed construction ever more closely, it means that this piece will be built, if at all possible, entirely with frame and panel work (rather than a corner-joined plank carcase as seen on the sideboard I made for the client last year). That piece combined carcase work with frame and panel elements. This time, I seek to only work with frame and panel elements. There’s one design driver all by itself.
The employment of frame and panel offers the chance to break the piece into discrete framed panel units insofar as possible. The doors are frame and panel units, as is the back of the cabinet. The top and bottom are similarly frame and panel, and then we have the latticed sides, which, though not paneled, are frames infilled with lattice and realize some of the same benefits as a frame and panel.
The biggest difficulty in designing with frame and panel is not so much in making the basic frame and panel units, for those are all much the same in how that are built. The difficulties come, initially, with how one connects the various frame and panel units to one another.
One choice for that is to use posts at each corner which connect the top and bottom frames of the cabinet together. Then the question becomes: how do we connect the posts to the frame and panel assemblies which (may) meet the post upon more than one face?
There are frame and panel units which attach to the post side-to-side, and there are other frame and panel units which will connect to the ends of the post. Each situation is a separate problem in itself, and each offers a limited range of solutions.
What becomes a most critical consideration in regards to the ‘how’ of the joinery, is that of ‘what’ the order of assembly needs to be. Order of assembly must be fully worked and out double checked, as you want to produce a cabinet in the end that can be assembled. I live in terror of being painted, as a self-inflicted injury, into that corner again.
Joining discrete wood frames to one another, or to posts or other intermediary elements, means that many of the connections are of a ‘sistered’ nature. If you want the parts to be closely together without gaps, or, to form a structural connection of some sort, and you want the joinery to be demountable and not involve metal fasteners, then the choices in solid wood joinery come down, as far as I know, to sliding dovetail keys, floating or incorporated in one of the members, or sliding hammerhead keys, or crosswise floating tenons secured with draw-bored pegs.
There are pros and cons to each approach, but one controlling factor is simply the question of how, and in what order, are the parts are to go together.
Sliding joint assembly means the use of dovetails, dovetail keys, or hammerhead keys, and those have characteristically two directions of movement, one in terms of how the parts are offered up to one another, and the second, generally at 90˚ to the first movement, in how the parts slide along one another to lock up the joint.
While those types of connections provide plenty of close lateral connection between the parts, the defect to the sliding connections is that they do not fix the joined assemblies such that they cannot continue to slide against one another from then on out. They might be limited to one direction of movement. And when they slide apart, they also tend towards greater looseness So, if you want the connection to be more rigid then you will have to add something to it to prevent the parts from sliding against one another. Not what you would call an insurmountable challenge, but when you start to pare the part dimensions down some, and things are getting crowded together at a confluence of joinery, say, then it may become less straightforward.
If you employ the floating pegged tenons instead of sliding dovetails then there are consequences as well in terms of assembly directions.
With the issue of joining a post 90˚ (or other angle somewhat close to 90˚) to a frame and panel assembly via the post’s ends, typically means the use of mortise and tenon. Connecting the post’s end to the frame is going to be done either at the corner of the frame where the joints are located, or some distance removed from that.
In this cabinet, the desire to maximize interior room for the overall volume, for one thing, precluded against the posts being setback from the corners of the cabinet to any significant degree, though pushing the pasts back away from the corners does open up options in terms of the joinery to be used, both at the corner and where the post meets the frame.
I needed to accommodate the front doors of the cabinet at the location of the front posts. The door hinging is configured so the doors can swing 270˚ open, and the profiles of the front posts and the hinging stiles of the door are mated closely to one another. I think this arrangement is vastly preferable to having doors which stop in the open position where you end up with the door frame free to run into surrounding support framing as it reaches the limit of its opening. It is the sort of accident which stresses the cabinet joinery and hinges, possibly dents pieces of cabinet woodwork, fails to win the heart and mind of the most ardent client, and the open door remains a continued hazard to walking in any case. Best avoided if you ask me.
In this cabinet, the front posts are to be shaped to accommodate the 270˚ swinging doors, while the rear posts are shaped so as to provide connections for the demountable back panel unit, a piece that normally is not opened or removed. Both posts are expected to connect identically to the top and bottom cabinet frames and maintain symmetrical reveals to those frames.
With the connection between post and top and bottom frame/panel units at the corners, and the top frame molded on its outside face as it is in this cabinet, certain joinery options are thereby manifest, and certain other options are not going to prove suitable. And as noted earlier, the deciding factor of which joint or even constructional system to use depends a lot upon the assembly sequence. It’s like understanding an interlocking stick puzzle, or burr. There’s a certain order to how things go together/come apart and if you don’t pay very close attention to that when you are designing then you could end up in a predicament.
Generally the joint which will work at the top corner is also going to work for the bottom sill frame/panel assembly corner. If I wanted to also keep a consistency in the look of the most exposed-to-view-joinery, then that aspect might be something to which I paid heed, but with the top of this cabinet the post tenon could be through and thereby exposed to view, while on the bottom of the cabinet the tenon end was not going to be visible even if through. So, that gave some leeway in terms of the details of the connections at top and bottom not needing to be the same as one another.
The joint between the post and the upper frame involves a tenon, which, if I were employing glue, could be kept relatively short. Here, with no glue to grip it, it needs to be a bit longer to accommodate some sort of mechanical stopping/locking mechanism. Given that the upper frame and panel unit joins to the post tenon as a pre-assembled component, then tenon, in order to be accessible for mechanical locking, needs to pierce right through the joint.
Now, the tenon could be fox-tailed, a concealed internal one-way wedging of the tenon otherwise known in Japanese as the ‘Hell tenon’. However, I want that demountability between the top framed unit, the post, and the bottom framed unit, so no use of hell tenons seemed worth considering. If demountabiliy was unimportant otherwise, then it would be a connection to consider.
Obviously, even in no-glue construction one could make exceptions for impermanent bonding agents, like a weak starch glue of some kind, or some wiped-on lacquer, to lightly reinforce a connection, which otherwise is fully dependent upon a tight fit between mortise(s) and tenon(s).
On the top frame of this cabinet the joint is fully exposed to view, while on the bottom frame it will be completely hidden. This difference leads to the possibility of different outcomes in joint design for those two places.
On the top, I choose to show the top of the tenon, but it is neither wedged nor is it reinforced with a weak glue. How then is the top frame and panel assembly prevented from being pulled apart from the post, you might wonder? Just m&t friction?
My solution was to make it a good friction fit joint, yes, and to also employ the side frames as an intermediary connection between the posts and the top frame. The side frames with lattice connect to the top frame’s end pieces with sliding dovetail keys, and the side frames are also connected to the posts with sliding dovetail keys. What prevents the connection to top frame from continuing to slide after assembly is that the flanking posts tenon into the same frame member. That blocks the side to side movement. The up and down sliding of the joint between lattice frame and post is blocked by way of a pair of wooden tapered pins.
The assembly sequence is therefore one of connecting side panels to the top frame end members, and then connect the posts to the side frames and slide the posts along the lattice frame until their top tenons pierce through the top frame’s corner joints.
For the bottom frame connection to the post, the only condition which is different, as previously mentioned, is that the post tenon which pierces the frame corner joint emerges completely out of view under the cabinet. So, I do have the option there of making a direct mechanical wooden connection, instead of borrowing the connection form that I did with the join to upper frame corner. Moreover, I am needing to do a direct locking connection to the tenon as a result of the previous decision about how the cabinet parts assemble to one another starting at the top. Once the posts have slid into place in the top frame, sistered to the side panels with lattice, then the option is no longer there to do a similar connection with the bottom frame end pieces and the bottom of the side frame as I had done on the top of the post.
If the cabinet bottom was the last frame and panel assembly, then the post tenon would need to either be trimmed flush with the bottom face of the frame member where it emerges, or some point short of that face.
If, however, the bottom frame and panel is not the bottom of the cabinet, if there are feet or some sort of sill assembly that raise it up off the floor, then other options arise for the post tenon joinery. The most obvious of these being that the post tenon can be elongated well beyond the face where it emerges, and is wedged crosswise. This approach is also preferred generally by me as it is a demountable connection. So, that’s the direction I tried to develop here.
This design direction initially did not come from trying to solve a joinery issue – I thought the cabinet should be a little bit more than the bare minimum of a simple box, and thought it would look better with some sort of plinth or base below the bottom frame, something to set the box on a pedestal.
Initially, I drew the cabinet with a supporting 4-sided skirt of sorts which joined at the corners with a miter:
The skirt had a surface which was concave, beaded, and which featured a surface jogging both inward and upward. One thing that could be said about it is that it was not going to be an easy component to make.
When it came to resolving the design details at the cabinet corners however, I tried a variety of approaches to satisfy the aesthetic desired. And yeah, it can be done in various ways. But, I wasn’t quite satisfied with any of the solutions I arrived at technically, especially when it came to figuring out how to terminate the pesky post tenon connection behind/into the skirt/sill and tied up in this puzzle was also the issue of how to attach the skirt itself to the cabinet’s bottom frame and panel assembly.
At one point, in working to solve this cluster of challenges, I redesigned the skirt so as to be composed not of four sticks joined with miters at the corners, but of four pre-assembled ‘miter corner units’ and 4 interconnecting stretchers between those corner units. This is an overall increase from 4 parts to 16 parts for the sill/skirt though.
This different construction gave the corners more of a look, it seems to me, of being upon pedestals:
The inner returns on the skirt could be made to curve inwards instead of outwards as they are shown above, and the pedestal look for the area would be replaced by a look of 4 squat sorta feet poking out.
Here’s a look at the underside of one of those pedestals, or corner units:
The triangular plate stiffens the corner and attaches to the corner with a tongue and groove. There are 2 floating tenons between the corner of the skirt and the lower frame corner as well to reinforce it. It was a workable solution but it still left me feeling like I hadn’t quite solved the problem yet, and I found some of the joinery on the awkward side. Despite the reinforcing steps, I also was left with the niggling feeling that the joints overall had a degree of fragility which led me to realize that I wasn’t quite done yet with the design.
I didn’t like the recourse to gluing of the corner miter, regardless of what sort of join was selected (through dovetail, concealed dovetail, etc.). I wasn’t too psyched either by the joinery situations which presented themselves at the other ends of the corner unit pieces, which were comprised of return pieces and then, close by, the off-center connection to the stretchers. These connections also seemed like a route which would involve the use of glue to effect a good result.
The assembly sequence dictated that the corner blocks go on last, and help to lock at least two of the stretchers in place, so there were concomitant issues with how to attach the stretchers to the bottom frame members. Again the problem arises of due to the narrow sistered joint options when eschewing glue in putting two sticks together.
The above solution – 16 parts with corner blocks, etc. – I ultimately rejected for reasons of unnecessary complexity. Some readers may guffaw at that statement considering the nature of the work of which I am generally preoccupied, and I, well, will just move along.
Nevertheless, further consideration led down a different path. I began by looking at the whole design which had started with the (original) considerations of ‘how should the piece look?’, and those considerations had brought me to produce a design with a mitered skirt of a sill.
That initial conception of a mitered skirt-sill was arrived at before I had fully considered all the particulars with the connections between the posts and other parts of the cabinet. While I had the top end of the cabinet’s joinery sorted out, the same had not yet been arrived at with the last bits of joinery at the bottom of the frame.
Now, however, the remaining issue of how to cleanly resolve both the corner post tenon joinery was first and foremost in consideration, and after considering again whether the skirt/sill should even be there or not, decided once again that I was keeping the something below the framework of the cabinet to meet the floor. A sill or plinth has the function of keeping the important parts off of the ground, besides any aesthetic role it may play. As it turns out, it also has a critical joinery role in this piece.
Once the joinery and assembly peculiars drove the design more than they had done in the initial stage, I moved away from the mitered ‘skirt’ idea to one of a half-lapped (looking) sill frame with some relief work applied:
More on the sill frame and bottom frame joinery will come in later posts, but for now hopefully a worm’s eye view will suffice:
With this support element, I have retuned to relative simplicity in terms of only requiring 4 main elements for the framework, rather than 16 as with the previous version. The emphasis on the lapped corner block areas is the visual similarity I think it forms with the preceding pieces I built for the client, both of which feature lapped pillow blocks in the support structure.
I didn’t rest there though, as I considered the above lap joint further, refining both the way the noses are cut and changing the type of corner joint employed, moving away from a lap, to obtain a more robust connection, now employing 8 parts instead of 4. A sweet spot perhaps in the part count? More on that part of the cabinet in a future post.
Meanwhile, I have managed to get some sporadic shop time in over the holidays and have moved the bottom frame corner joinery along some distance:
A closer look at one corner, which will obviously need a few passes with a plane yet, not to mention some (previously-planned) chamfering:
I wish I had been able to have a piece of stock which did not have that small portion of wane on it, but that is not characteristic of the material I have on hand. The chamfer I will later apply should take care of it.
Another corner, another rotation of the camera:
I think these are some of the cleanest shachi sen mortises I have ever hacked out. Not sure quite what happened there, but I hope to continue in this vein with the project. The above pictures of the corners are of the initial fit up, without any tune-ups of any of the parts, if required. Somehow, it was good to go from the get-go.
Also there was some layout on the assembled frame for the post tenon mortises to take care of:
A fortunate thing happened there, in light of the discovery that the above mortise layout is significantly off at each corner, was the thought to cross-check the dimensions between mortises. Once the discrepancy from what the numbers should have been was revealed, I regretted the ink dots on the above layout, but far better to have that to deal with than having knife lines incised on there in the wrong place – – or having not done the double check measurement and gone ahead and cut the mortises, only to find that they were out 1/8″ (3mm) in both x- and y-axis from one another. That would have been a most unfortunate outcome. So, whew!
Sorry for the long winded post. I hope it was interesting at least in parts, and we’ll see you back around here some other time. All the best for 2019!
Next up in this series is post 14.