Issue 290 of Furniture & Cabinetmaking is now in print, and includes the next article in my Roubo bench series, which this month is all about processing the monolithic slab top by hand.
Issue 290 of Furniture & Cabinetmaking is now in print, and includes the next article in my Roubo bench series, which this month is all about processing the monolithic slab top by hand.
I’m going to be honest with you, dear reader. Flattening the bench top has been the single most challenging aspect of the Roubo build. Not the large-scale joinery (which is where I expected the challenge to come), but getting the bench top flat and true. Why is that? I flatten every component of every build. And while that may be true, at 102″ x 21″ the bench top is the single largest surface area I’ve tackled, and the only project which has truly dwarfed my big No8 jointer plane. That being said, with perseverance and trusting fundamental handwork techniques, even a large surface area like this can be tamed.
At the end of my last blog post the back half of the slab had been traversed flat across the width. Since then I have traversed the front half, and then did some measuring with the straight edge to build up an understanding of the topography of the slab. A couple of truculent knots, and some pronounced cupping, meant that the slab had quite a lot of variation before being traversed. There was also a stubborn low patch along the rear edge, between the two legs.
After traversing, the next step is to work down the length on a 45 degree skew, working in one direction and then the other. The basic premise is that each process removes the variations of the one before, finishing with jointing along the length. I started working on the skew, and found myself chasing my tail, introducing a fall-off on the far edge where it had previously been flat, and generally making no progress. That’s ok – it’s wood and it hates you, and sometimes things don’t go quite to plan. My preferred solution in such circumstances is to take a break, preferably for a day or so, and return to the task fresh. I also took the opportunity to correspond with Mark Hicks, who kindly gave me some helpful pointers on how to tackle flattening such a monstrous piece of oak.
My first step when returning to the workshop was to sharpen my No5 and No8 planes, and to undertake a hard reset of the slab by traversing it flat again. This didn’t remove too much material, but did give me a solid reference surface from which to work. One of the interesting questions with flattening benchtops is “how flat is flat?“. You can work until you’re getting full width shavings off the plane, but after my recent frustrations I decided to be a bit more scientific about the process. So, I dug out my feeler gauges and traversed the bench until the entire top was flat to 0.005″ over a 24” length. I found that marking the high spots with a red lumber crayon helped to identify the key areas to work.
Once the benchtop was traversed flat, I went back to working at a 45 degree angle with the jointer. This can be an awkward stroke as the plane ends up being extended far in front of you, which for the heavy No8 can introduce a nose-dive, rounding over the far edge of the workpiece. Paying attention to body mechanics, and being careful to end the stroke with pressure applied to the heel of the plane, is essential. I also measured the bench top every few inches with my straight edge and feeler gauge, marking off the high spots to ensure that the plane was removing material where it needed to. After six passes down the length of the bench top in each direction, the surface was flat to 0.005″ over 24″ on both the diagonal and across the width. Which seemed like a very good place to stop for the day. The length is a bit lumpy, but not more than 0.008″ on 24″ for the majority of the surface, and jointing will remove that in relatively short order.
Flattening the benchtop has been humbling for sure, but moving through those difficulties and getting it to a state where is pretty much finished is very gratifying.
The latest issue of Popular Woodworking landed on my doorstep yesterday. I’ve been looking forward to this particular issue because it includes an article I researched and wrote at the start of the year about the restoration of the iconic Charles Rennie Mackintosh designed “Willow Tea Rooms” in Glasgow.
Following PopWood’s entry into a Chapter 11 bankruptcy protection earlier this year it wasn’t clear if this article would ever see the light of day. Fortunately, with Active Interest Media’s purchase of PopWood, the magazine seems to be in better shape.
I’m grateful that this article is now in print, as it examines a painstaking (multi-million pound) restoration of a hugely important Mackintosh landmark, and shows the very finest of contemporary Scottish craftsmanship. It is a breathtaking project, and I hope readers will find it as interesting as I have.
With the Roubo bench now in position in the workshop, the finish line is starting to come into focus. That being said, there is still plenty left to do, including flattening the bench top, shaping the vise chop, cutting the planing stop mortise (and fitting the planing stop), abd drilling the holdfast holes. I’m hopeful that the bench will be in use by the end of the year, even if there are a few things left to finish off in January.
The first thing I did to the assembled bench was to trim the drawbore pegs flush on both the external and internal faces of the legs, cutting the waste off with a flush cut saw. Flush cut saws always feel like a compromise tool to me – in theory they should do the job perfectly, but in practice it is very easy to find them marring the surface of the work. To avoid this, I gently angled the saw just a hair away from the surface of the leg. On the show surfaces I then pared the remaining waste flush with a paring chisel. On the internal surfaces of the legs, I left the pegs as they were cut off the saw.
Now that the bench was looking less like a well staked vampire-Roubo, it was time to start flattening the bench top. This will need to be done periodically as the slab moves, although my expectation is that after the first year or so the slab will move very little. The first flattening takes the most work, as the top side of the slab had not been dressed before assemling the bench. This is critical work, as I want a flat surface to work on free of twist or undulations, and so I decided to pace myself across several sessions rather than rush to get it done in one go.
The first step was to get an understanding of the topography of the bench, using winding sticks and a 60″ straight edge to identify where the main bumps and hollows were. The top was roughly “m” shape in cross section, with low points along each edge, and two bumps separated by a hollow in the centre of the top. Traversing the top with a jack plane (my normal method for heavy stock removal) is ideal for getting the bench close to flat, as the plane will skim the tops off the bumps and bring them down to the level of the hollows. I divided the bench length into two sections, and focused my initial efforts to the rear half.
Before traversing, I decided to flush up the end grain of the leg tenons, as these were protuding through the bench top in a number of spots. That sheer mass of end grain is not much fun to plane with a bevel-down hand plane, although it can be done. Fortunately, I had recently invested in a Lie-Nielsen No62, which as a bevel-up plane works wonderfully on end grain, and flushed the end grain rapidly leaving the bench top ready for traversing with the Clifton No5. The No62 is not essential for this task, but having it to hand did make tackling the slab much more pleasant, including on some harrowing knots which the low angle blade geometry cleanly slices through where my No5 wanted to ride over.
I left the bench top with the back half traversed flat. When I am next in the workshop I will traverse the front half flat, following which I will then joint it so that there are no bmps or hollows along the length.
Glue-up is often the most stressful element of a build, even when it goes smoothly. So glue-up of a piece that cannot be easily disassembled in the event that something goes wrong, and in the presence of other folk? I’m going to be honest and say that I wasn’t much looking forward to assembling the Roubo workbench. That being said, all of the work fussing the fit of the joints, and mentally rehearsing the sequence of assembly countless times, paid off. The glue-up went without incident and I owe a debt to my team of helpers.
The set-up for assembling the bench was quite simple. I had three portable heaters running for a few hours before hand to get the temperature to hide-glue friendly levels (I used Titebond Hide Glue for this assembly as it has a lower working temperature than Old Brown Glue), a bucket filled with boiling water to keep the glue bottles warm and to help with clean-up, a pair of lump hammers, and solder brushes for glue spreading. No need for clamps for this glue-up!
The first stage of the process was to glue and assemble the undercarriage. We made two sub-assemblies each of a pair of legs joined by a short stretcher, then inserted both long stretchers into one sub-assembly and manouvered the second sub-assembly onto the opposite end of the long stretchers. At this stage the stretchers were not clamped in, nor were the drawbore pegs inserted, as I wanted to allow the undercarriage to flex a little when the legs were inserted into the top. After a quick clean up of squeeze out with wet rags and toothbrushes, we then painted the double tenons on the legs with glue, and moved the slab top onto the legs. This was the moment I have been simultaneously looking forward to, and dreading. The slab settled nicely about three inches onto the tenons just under its own weight, and gentle hand pressure moved it slightly further along. To drive the tenons the rest of the way through, we lifted each end of the bench (by the short stretcher) in turn, and dropped it from about 6″ off the ground. The shock of the legs hitting the floor drove the slab deeper onto the tenons, and after three drops per end it was fully seated.
As a final step we then inserted the drawbore pegs to achieve a completed clamp free-assembly. The pegs were covered in glue and then driven home with a lump hammer, after which we did a thorough clean up of squeeze out and drank a celebratory beer while feeling quite satisfied with life. The fit of the tenons in the top looks ok, and there were only a few gaps big enough to drive wedges in. That may change as the bench settles and I flatten the top, but I have plenty of wedges prepared (all cut from an offcut of a stretcher) so can gap fill as necessary.
The workshop was feeling quite cramped wih two benches in situ, so a few days after we had assembled the bench I reconfigured the ‘shop layout. The Sjoberg bench is now at the end of the ‘shop where it will serve as my shapening station, replacing the temporary sharpening table I set up when I moved into the workshop 5 years ago. With the Sjoberg relocated, I was able to move the Roubo into position 6″ out from the left hand side wall, and re-hang the saw cabinet on the wall. It’s funny, but now that the Roubo is in position it looks a lot less monolithic than when I was working the slab over the past few months. At 102″ long it is going to be a very useful work surface, and I can’t wait to finish the bench and start making furniture on it. The next tasks will be to trim the drawbore pegs flush to the legs, and then flatten the top,. Roubo is not here quite yet, but the end is now in sight.
I find that most projects have a reassuringly familiar tempo. Stock preparation and the early stages are quiet leisurely, the pace quickens a little as you get into the interesting joinery and decorative elements, and then at a point just before the main glue-up everything starts to speed up. For the Roubo build (which started in February) this has held true, save for a two month break over the summer when I was working on my stick chair for the Mortise & Tenon article. The increase in pace just before assembly has definitely been felt on this project – due to the sheer weight of the slab I’ve needed to draft in a team of helpers when it comes to assembling the bench, and given how busy everyone gets at this time of year we had to schedule the glue-up date a ways in advance. We’ll be assembling the bench tonight (Roubo is nearly here!), and as a consequence every available moment over the past few weeks has been spent getting ready for this deadline. As is often the case, you think there is not too much left to do on a project and then you remember all of the small tasks necessary before assembling it for the final time. And so my evenings have been very full (and often late finishing).
The biggest single task has been to tune the fit of the legs in their respective sockets. If you have not yet watched the Lost Art Press video on building a Roubo bench, I highly recommend it. That video, and Chris’ book on workbenches, have been invaluable resources for this build. There is a moment in the video when Will Myers tests the fit of one of the legs, and it casually slides 3/4 of the way into the socket, coming to a final rest with only a few taps of the mallet. Well precisely none of my legs fitted like that, which I was expecting (still, it would have been nice). So I’ve been tuning the fit, making sure each leg moves nicely into position but is also good and snug. I think this work has partly resulted because I approached cutting the mortises as I would cutting furniture joinery – cutting against the line rather than on it, and aiming for a super snug fit. And actually, I suspect that a Roubo work bench needs a little slack in the joinery to allow it to all go together. Not cavernous gaps, but slightly more slack than you would aim for in fine furniture. Or maybe I’m just trying to make myself feel better.
In any event, some tuning has been necessary. This is a case of dropping the leg into the mortise, looking for the burnished areas on either component which indicate a too-tight fit, and gently easing those out. On the leg, I’ve found that refining the external surfaces with a smoothing plane provideds a very controlled method of fine adjustment, while the internal surfaces of the tenons need help from a paring chisel, as do any high spots in the mortises. Here’s one of the odd things about writing about woodwork – you can summarise a process in two sentences which might actually take hours of painstaking work. Easing the fit of the legs while trying to avoid introducing horrid gaps in the joinery meant that this was fine and pernickety work, taking on average two hours per leg. But it was time well spent, and the end result is legs which drop about 3″ into the socket under hand pressure, and which then drive home to 1/4″ shy of fully seated with persuasion from a 1lb lump hammer. I have resisted the urge to fully seat the legs for fear of extracting them – the legs are pretty dry and this stage, while the slab will take decades to dry out. Moisture transfer from the slab to the leg will help to lock it place, which is good at final assembly but far from optimal at this stage.
Building a workbench is an exercise in creative workholding, particularly when tuning the fit of the legs because the slab is on top of my Sjoberg bench. To facilitate easy working while tuning the fit of the legs, I clamped a Bessey K clamp to the slab, and used the clamp head as a makeshift planing stop, which worked quite well.
Other work included checking the fit of the stretcher tenons, and tuning these where necessary so that the stretchers drop into their mortises under hand pressure only (the joinery for the stretchers was cut in March, and we have had a lot of humidity changes since then). I’ve also chamfered the corner of the underside of the slab top, and the corners of the legs and stretchers, all with my new Philly Planes chamfer plane (a luxury item for sure, but one which makes cutting consistent chamfers a breeze, and Phil’s customary crisp workmanship means that the plane is as lovely to use as it is to look at). This was only a light chamfer, but will protect the delicate edges in use. The feet were also chamfered, this time with a 9 grain rasp, to avoid spelching the grain when moving the workbench.
Boring the draw bore holes and shaping pegs was the final task. I used straight grained, riven oak stock, to ensure that the pegs are strong with grain that runs consistently from tip to tip. The pegs are shaped slightly oversided using a block plane, and then cut using a Lie-Nielsen dowel plate to tghe final 5/8″ diameter. When planing the rough pegs, I find that a bench hook provides a good way of holding the workpiece.
Today I will be doing final prep for assembly, and keeping every available digit crossed that we have a smooth glue-up – this is not one where taking everything apart to ease the fit is really possible! But I am looking forward to it, and to hitting a major milestone on the bench build.
Today is a momentous day – the joinery for the Roubo bench is now all cut. The one exception to this is the mortise for the planing stop, which will wait until the bench top has been flattened. But right now, all of the structural joinery is done, which feels like a huge milestone.
Cutting the dovetail sockets for the legs is much quicker than cutting the mortises, but in many ways it feels more nerve wracking. Cutting a square mortise to the right dimensions is relatively straight forward, all told. But cutting the angled shoulders for the dovetail sockets – critical joints which will be visible every time I step up to the bench. That feels pressured. As often is the case, in practice it was not as tough as I had expected, although I am glad that I started with the rear pair of sockets first (these will face the workshop wall, so I won’t see them very often), to warm up. Breaking the operation down into a clear set of stages helps, as does remembering that joinery like this is just a series of fundamental hand skills (accurate layout, cutting to the line, and some chisel work to remove the waste).
First I transferred the layout from the underside of the slab (where I had previously traced it from the legs) to the top of the slab. This is simply a case of taking the precise angle of each side of the dovetails, and depth of baseline, from one side of the slab and striking corresponding lines on the other side.
After a bit of experimentation, I found that it was easier to cut the angled shoulders with a coarse cross-cut hand saw rather than a back saw, and my Skelton Panel Saw made short work of this critical cut. Setting the bevel to the right angle and standing it a few inches from the cut provided a clear visual guide as to how far to angle the saw plate. Cutting joinery with a hand saw feels counter intuitive at first, but works very well. I started each cut at the near corner, where I could see both the line across the width of the slab and also the angled line on the face of the slab. I knibbled a saw kerf along these two lines, and once I had hit the baseline of the angled cut, and the far corner of the straigh line, I allowed the toe of the saw to drop, taking full length strokes of the saw until the cut was complete. This is very much how I cut tails for furniture-sized dovetails, just on a much larger scale.
To remove the waste I also cut five relief cuts in each socket, and then knocked out most of the material with a 1/4″ mortise chisel and mallet, working from each side of the slab into the middle. To avoid bruising the interior of the socket walls, I removed the waste from the middle of the socket first and then cleaned up the waste at the edges. The waste pops out easily, making this a very efficient way of hogging out a lot of material.
To take the sockets to final depth, I use a similar approach to how I cut the mortises. First I deepen the baseline with a 2″ chisel and sharp tap from a mallet, followed by a 1 1/2″ wide chisel and mallet to get very close to the baseline. Once there is only a small amount of material left I moved to the big timber framing chisel. Although this chisel is huge, I find that it is very effective as a paring chisel when working across the grain, as the sheer mass means that it will cut without riding up over any difficult patches of grain, resulting in a flat bottomed socket. Ordinarily I would use a router lane for this task, but the sockets were deeper than my router plane could reach.
My next task, once my slab moving team have helped get the slab back onto my existing bench, will be to test fit the legs into their mortises and to tune the fit where necessary.