June 20, 2006 - The above chair is the subject of this reproduction process, which will be described as it happens in these pages.
Inspiration often pops up when you least expect it. Subject chair was purchased at a garage sale for ten bucks a few years ago and has been a mystery as to its origins and why there was only one.
An inspection of the chair reveals that it was made by the hand of a master craftsman (with an obvious love for what he wrought that shows right through the clumsy attempt of some misguided re-finisher), and, judging from the joinery and hand tool marks, more than likely around the turn of the last century (circa 1900).
The chair is white oak, with pegged mortise and tenons, all the parts are from 1" to 1 5/8" thick stock (no 3/4" stuff in this baby!) and it is as solid as the day it was built.
It is definitely not a beautiful chair to look at, but comfortable despite its plain looks, and its beauty is in the details of its construction, and the knowledgeable skill of the maker, which shines through like a beacon to those who know and appreciate such things.
Upshot is that the only thing I would like better in a chair to go around my own Mission Trestle Table, and the "Arts and Crafts Bench is six more exactly like the one above, in quarter-sawn white oak.
... and perhaps, at the same time, pay tribute to a woodworker from another era whose skill has stood the "test of time".
So, here we go:
The Planning of the Prototype.
Because of the many angles, some compound, the plan is to first do a prototype using a cheaper wood (poplar) to work out all the joinery details.
For this project a decision was made to use "loose tenon" joinery, a variation of traditional mortise and tenon joinery, whereby a mortise is cut into both pieces to be joined and a "loose tenon", milled separately from the same material and cut to length, is glued into the opposing mortises to join the two parts.
Research and previous experience indicate this variation of traditional M&T joinery to be sufficiently strong for the task. It also has an added benefit of allowing the various parts to be batch cut to final dimension, making for increased accuracy and more efficient use of shop time in a small production run that has the ultimate goal of producing identical parts for six chairs.
The first step is to take measurements of the original chair and do a full scale drawing on 1/4" MDF of one side of the chair frame. Since each chair will have two identical side frames, each with a front and back leg, a side seat rail, and a stretcher, this is as good a place as any to start. We'll worry about the parts that go between later.
The next step is to do a template for the back leg out of 1/2" MDF, which can be seen at the right side of the first picture above.
This template will be used with a pattern bit to identically route all the back legs of the chairs. I spent the better part of an afternoon dimensioning, cutting out, and fairing the curves and angles of this back leg template. Since each leg will be an exact copy, it pays to take the time and get it right.
While we're at it, and with the complete back leg template in hand, now is a prudent time to "dummy up" one front leg, one seat rail, and one side stretcher out of MDF in order to work out the angles and dimensions for the joinery. You can see this process in picture two above. This step will also provide dimensions for a good start on a cut list, which will be necessary to order the wood when we're ready to build for real.
With the back leg template complete and joinery locations marked, it is not a bad idea to make extra copies for safe keeping. Using the original to trace out two copies on MDF, rough cutting them on the band saw, and using a pattern bit on the router table, produced two additional copies of this all important template for when Murphy strikes.
At the same time, some 8/4 poplar was milled, then routed using the same process to produce the two back legs for the prototype.
Clamping them to the master back leg template at the same time makes it easy to precisely transfer the joinery locations onto a matched pair of legs ... and now the parts are starting to resemble one side of a familiar chair frame.
As anticipated, working out methods to cut the project mortises is turning out to be time consuming in the early stages, and it's easy to see why so many small shops are going to the JDS Multi-Router to simplify the task.
Besides the fact that routing angled mortises into the faces of legs with this particular homemade jig has a great deal more imprecision built-in than anticipated, it also takes four separate, carefully crafted, "shims" just to cut the four joints, with three different angles, shown above.
Finding router bits long enough to do the job with this jig has also been a problem, solved with the use of extra long, carbide, "end mills". The 3/8" joints above were cut with multiple passes of a 1/4" carbide end mill, and, as a result, sizing the loose tenons is problematic because the shoulders can't be easily rounded over using a standard round over bit.
I am not really satisfied with the fussy nature of the setup, so it's time to rethink the basic jig, and re-build it for a 3/8" end mill that can do the proposed 3/8" mortise width in one pass, which will hopefully solve the imprecision noted thus far, as well as the problems fitting the loose tenons. Back to the drawing board.
The solution to the jig problems on the project thus far can be seen in the background ... the JDS Multi-Router. In the short span of a few hours, including coming up to speed on a new piece of equipment, we now have a prototype chair frame, made up of numerous compound angles, all done with floating tenon joinery and using the Multi-router to cut the mortises.
While there is still some evidence of imprecision due to two separate methods of cutting the mortises, the light at the end of the hoped for, small "production run" tunnel is apparent. With the introduction of this one tool, batch cutting, and the subsequent routing of parts should improve the overall precision and make this project move forward with much greater speed
Jigs for jigs ...
The next challenge to be met is to come up with a "production" method for doing the 12 mortises needed for the six backrest slats in each chair's curved crest and intermediate rails.
A total of 12 of these curved rails will be needed (two for each chair), not counting extras, so the ability to do these 72 mortises quickly and accurately is important. (Note: the intermediate back rail pictured above is just a spacer)
To the left is an mdf prototype of the "crest rail", looking at it from the bottom side and showing the desired location of the backrest slat mortises. The second, "intermediate" rail's mortises will be in a mirror image location on its top side.
In a sort of "belt and suspenders" approach, two jigs/methods were simultaneously developed. The first centering around the Multi-router, and the second using a plunge router to perform the same task.
Method 1 uses the Multi-Router with an additional jig:
Since the curve is an arc of known radius, and the center line of the mortises must be perpendicular to the "chord" of the arc at each mortise location, it was a logical jump to use these elements in designing a jig to be used in conjunction with the Multi-router (just think of Samuel Colt and the development of the "revolver", which works on a variation of the same principle):
The jig is first referenced, then clamped to the center line of the x/y table. The rail to be routed is then clamped along the curve of the jig body, which corresponds to the same arc radius as the rails.
In operation, each desired mortise location is subsequently revolved along the arc to the top reference line of the jig, where it is re-clamped, then routed precisely perpendicular to its chord in the manner in which the Multi-router excels.
The slot in the jig front also aids in referencing the router bits travel, and, used in conjunction with the Multi-router's built-in stops, makes it easy to set the mortise length. The "depth of cut" line can also be seen in the last photograph above.
Method 2 uses a plunge router with a "collar guide" equipped base.
The idea for this jig is based on an article by Terry Moore in the March 1990 issue of Fine WoodWorking magazine, entitled "Production Chairmaking".
These two pictures should be self-explanatory. The workpiece is simply inserted under the top between a curved plate and a cauls and the whole enchilada slipped into a bench vise.
Both jigs/methods are simple solutions to the problem of locating, and quickly and accurately routing mortises in these curved chair parts.
Method 2 has the advantage of being within easy reach of those without access to a Multi-router or slot Mortiser (although in this case I did use the Multi-Router's capabilities to great advantage in locating and routing the initial collar guide slots for the jigs top).
Method 1, besides taking advantage of the unique capabilities of the Multi-Router, has the ability to easily change the locations of the mortises by simply marking the parts appropriately, without having to "re-invent the wheel" for different chair back designs. Once made, this jig is immediately ready for future chair projects and it greatly enhances a tool already superior at tasks of this nature.
Although there are undoubtedly many alternate methods, either of the above jigs handily solve the last big challenge in finding methods to do this small shop "production run" of chair parts.
I still have to finish the jig(s) for actually making the curved crest and intermediate rails, but this is the same, time honored methodology (using a router and pattern bit) that was used to do the back legs, so it is familiar territory.
Above, the chair frame, dry fitted, with the joinery details worked out and complete.
All that's left to fabricate is the seat, but we first need a stable, frame to work out those details, so it's time to think about assembling and gluing up the prototype's frame.
First step in a multi-step glue-up process, and an excellent opportunity to rehearse and get the sequence down for the real thing.
In most complicated, multi-part glue-ups, a little thought will suggest a sequence that will allow for ample time, and one of the first things is to decide whether there are any sub-assemblies which can be glued up first.
In this case the backrest, with the crest and intermediate rails (above left), and the slats; and the three leg stretchers (above right), are two logical candidates for sub-assembly glue-up.
A caveat: when gluing up sub-assemblies, and after the glue and clamps are applied to same, it is not a bad idea to go ahead and dry fit and clamp the rest of the components to the sub-assemblies to insure that everything fits as it should ... although not pictured above, I usually leave the whole enchilada dry fitted and clamped together while the glued up sub-assemblies dry.
The prototype's frame, fully glued, standing on its own legs, and next to it's ancestor as a likeness check.
Wood, or upholstered seat?
I'm not a fan of solid wood seats in a dining side chair and after spending this much time on the project, and even though the original intent was to faithfully reproduce the original chair first shown above, I hate to miss an opportunity to try a method that may possibly work better for the overall design. After all, it is a prototype and experimentation is allowed.
So, for now, at least on the prototype, we're going to experiment with a traditional upholstered seat over webbing stretched across the seat frame, shown above, which is more comfortable than foam on plywood:
This method is also traditional and typical of many Harvey Ellis and Gustav Stickley chairs of this period, although they used horse hair as the padding.
The next step is to put some type of finish on the prototype and put it into service, both to see how it stands up to actual use, as well as to give the upholstered seat a trial run while we build the frames for the production run of six chairs.
As it turns out, and even though it was designed as a prototype made of cheap wood, we decided to keep it. So here is the finished prototype, now a keeper, with one coat of General Finishes' "Java" Gel Stain, and four coats of sprayed shellac.
Now, enough of this prototype business ... onward to the real thing.
NOTE: After much angst/discussion it was decided to forego the upholstered seat for the final six chairs and maintain the spirit of the original wooden seat with the following:
10/20/06 - The end result of the "reproduction" process, with one of the six chairs fabricated from lessons learned during the prototyping, shown next to its parent:
10/22/06 - And here are those six chairs, the purpose of this entire exercise, completed and shown with the Kitchen Dining set: