You are currently browsing the tag archive for the ‘woodworking’ tag.

20170627_210550_New Quay Court

Finished posts, coated with tung oil

Well progress is definitely happening.

20170622_185235I cut out my blanks for the samson posts on the bandsaw.

20170625_184644_New Quay CourtThe time consuming part was shaping the top of the post. I had made a pattern for this.

There are a variety of ways of cutting a concave curve. The method I chose is essentially the same as making a housing, but using a gouge rather than a chisel.

20170626_170958_New Quay CourtHaving marked the pattern all round, I cut down to the lines with a tenon saw. Then I cut out the waste, trimming carefuly down to the line. Having down this on two opposite faces, it was simply a matter of removing the waste between the finished lines.

The second pair of curves were effectively marked with the saw cuts. So I just drew in the curve and repeated the exercise.

The rest of the process was the same basic technique as spar making. You turn the square octagonal, then 16 sides, then sand smooth with 60 grit paper.

Of course it all takes time, and care, smoothing out bumps and getting curves that feel good. But really, like so many things, the path is simple if it is done in the right order.

Finishing off I used my favourite tool the cabinet scraper, which saves hours of sanding.

I had to glue an extra piece on the flare at the back, because the timber stock I had available was not quite wide enough. This was only a thin taper. But it is worth mentioning that when this kind of thing happens, the extra piece should be kept chunky and oversize. It is much easier to glue on a decent thickness of wood, because it remains stiff. If I had shaped the thin taper first, it would be very hard to glue it on and would bend all over the place. Glued on oversize, it was just a matter of planing flush, band sawing off the excess, and planing the taper.

The finished posts look well. I have oiled them, and tomorrow I will dry fit them. Then they will be varnished before going in place.

20170627_210550_New Quay Court

Finished posts, coated with tung oil

20170626_121102_New Quay Court

The  knock down portable work bench

One of my favourite principles in doing pretty much anything is what I call ‘circuits’.  It’s all about getting the energy you are putting in, precisely to the place you want it.

Anyone familiar with metal cutting equipment like lathes and milling machines knows that a great deal depends upon stiffness.  The tool and the workpiece must be held rigidly.  When the tool is presented to the workpiece, any ‘slack’ in the physical path between the tool holder and workpiece will be taken up before cutting happens, causing wasted energy, inaccuracy and vibration.  The ‘circuit’ between the two must not have any energy leaks.

The metal worker, operating at thousands of an inch or less, knows that these ‘leaks’ are visible enemies, and will destroy his work.  They will cause the workpiece to deflect, set up vibration patterns, even cause ‘dig ins’ as the vibration pattern is superimposed on the force attempting to cut.

The woodworker is operating at bigger tolerances, so this problem is not so obvious. When a piece of timber moves as it is sawn, it is often just accepted.  When it bounces as it is struck with a chisel, it is accepted.  When the workbench wobbles as a piece is planed, it is accepted.

Well, not by me.  The point is that *any* slack in the system creates an enormous loss of power and accuracy.  I don’t really think many people realise how much energy is used in moving stuff around, rather than cutting, and what this actually means.

For example try this experiment.  Secure a piece of 1″ x 1″ wood in a vice so it is sticking out about 10″  Then try to cut it with a saw about 6″ from the vice.

You will instantly notice that the wood will vibrate as you cut it.  The energy you are using starts to fight against you.  The vibrating saw cut jams the saw.  It is impossible to cut a straight line because the wood always wants to bend.  The jamming of the saw causes the saw to vibrate and the wood to bend even more.

Now try the same experiment cutting very close to the vice.  All the thrust on the saw goes into cutting.  It is easy and accurate.

The difference between these two experiments is pretty obvious.  Cutting at a distance from the vice turns the wood into a spring that absorbs the energy of the cut, and then releases it in an unhelpful way.

If you properly secure a workpiece in an immovable vice, the effect is astonishing.  Cuts are effortless and take a fraction of the time. you can be stunningly accurate, paring off transparent shavings.  Everything is way, way better.

Securing your workpiece and cutting it close to the vice or clamp is second nature to craftsmen.  But how do you know your clamp is secured? What is going on between that clamp and the ground, and the ground and your feet? All you need is a slight wobble, a slight give in your bench, and all that sharpness vanishes.

A workbench is like a system of springs between you and the workpiece.  As you apply pressure to a workpiece the springs give, absorbing your cutting energy.  The more they give, the more energy is absorbed.  Only at the point that the wood you are cutting cannot resist the force of the ‘spring’, will the wood actually cut.

If there is any slack in the system, your workpiece will travel the distance of that slackness before it is cut.  So you are cutting a moving target, that springs back as you cut it.  The result is far lower accuracy and far less cutting power.  Even a small amount of spring makes a massive difference.

One question that came up some years ago was whether it was possible to construct a mobile workbench with very high stiffness, that could be taken apart easily.  Here is my solution, that is still going strong.  It could certainly be improved upon, but it is way better than most static workbenches, let alone mobile ones.

The basic principle that is essential in a mobile bench is that the circuit between the operator and the bench must be closed.  In practice this means that the bench must include a floor on which the operator stands.  The good old Workmate actually does address this, by having a platform you can put a foot on whilst using it.  But it is extremely springy, and the modern version is very poor quality.

My solution was to have a portable floor.  I made this from OSB, but I think a good quality plywood would be better,

The easiest way to describe it is to show you how I put it together.

Components are the floor, a base, three legs, the bench top/vice and three clamps.  The whole thing takes a couple of minutes to knock down and build, and easily fits in a car.  The bench part has only a single M12 bolt holding it together. The bench and legs are made from Keruing, a hard dense and heavy tropical timber, that I happened to have on hand.  I would recommend a heavy hardwood, because the mass definitely helps.  The base is softwood.

20170626_115524_New Quay Court

The components.

The base that holds the bench is made from two pieces of timber in a ‘T’ shape. The  upright of the T is dovetailed into the long part.  The ends of each part are mortised to take the bench feet.

20170626_115555_New Quay Court

Softwood base, dovetailed together

The bench itself is two pieces of hardwood, that clamp over the legs.  The legs are half-dovetailed into this, to that once the two pieces are held together, the legs and bench form a rigid structure.

The third leg is bolted through the bench.  This bolt is the only fastening holding the whole thing together.

20170626_120136_New Quay Court

The third leg.  This bolt is the only fastening!

Once the bench is fitted to the legs and the bolt loosely done up, the whole thing is placed on the base and the tenons on the legs slotted into the mortises.

Then the base is located on the OSB floor.  The floor has captive ‘T’ nuts underneath that take the bolts that go through the base.  The bolts pass through timber clamps that fit the angle of the legs, and effectively create a kind of dovetail effect.


The clamps bolts the bench to the base and create a dovetail effect on the legs.

This structure is amazingly rigid.  With forces along the vice, it is exceedingly stuff, better than most static work benches.  There is a small amount of lateral movement owing to the flexibility of the OSB but I have got around this by putting weights on the floorboard, or wedges underneath it at the back. An updated version might include stiffeners for the floor board.  I tend to use the centre of the vice for sideways loads, as the forces are resisted very well by the rear leg.

20170626_121102_New Quay CourtIn using the bench, one stands on the board.  Thus the circuit between the user and the bench is always fixed.  You can stand in front or behind.

The vice is fairly primitive, using captive nuts and 16mm studding.  At the time I had only basic metalworking equipment.  Ideally I would have sliding bars to prevent racking of the vice, and maybe handwheels.

Some kind of stop could prove useful. You could even put an end vice on it!

20170626_121722_New Quay Court

In action, turning a chunk of oak into a Samson post



20170618_164409_New Quay Court20170618_164405_New Quay CourtFor years my lovely gaff cutter has been hanging around next to our massive houseboat project, sadly neglected.  Time to find out what is going on.

I knew that the starboard side deck and after deck had rot, so off they have come, revealing the bones beneath.  It is a painful thing running a circular saw across your boat’s deck.

Well, it’s not too bad.  The eyebrows had to shoot up at the timber selection of some of the deck beams.  My goodness, some serious grain run out, to the point where a couple of the half dovetails have parted.  Still these will be fairly simple to replace.

A bit more thought will have to go into a section of the carline, which is certainly rotten at the top edge.  I’m not yet clear how far it goes, but it may be a case of scarfing in a repair.  I certainly want to avoid replacing sections of it.

The samson posts on the aft deck have decayed sufficiently to weaken them at deck level,  One problem was insufficient clearance between them and the transom, so water has been trapped.  I’ve made a new design that gives me  a bit more clearance.


The timber I have to hand does not quite fit the new pattern, but I’ve found some that’s close enough and I can get away with a small shim to make up the extra width.

It’s always fun searching around for timber that fits a shape, and then revealing the inside.   This piece is a 5″ thick board of oak I have had hanging around for about twenty years, and the grain matches quite well to the angle between the sloped transom and the more vertical post.  My lovely Sedgewick planer and bandsaw will do the rough work…20170622_185235The worst rot revealed by lifting the decks was unexpectedly in the port quarter block, a sizeable chunk of timber that connects the beam shelf to the transom, provides corner strength and holds one of the large iroko davits.  This is sufficiently decayed to merit complete replacement, so I’m fishing around in the wood pile for something that works.

Deck plywood has arrived from Robbins, and bronze nails, so I’m really looking forward to quick progress!

We’ll see…

In accurate large scale work, long straight edges are often essential.  But how do we get to an edge that is really straight over a long distance?

We can use chalk or string lines or lasers, but here is a low tech method that does the job very well and quite fast.

This came up in a project involving constructing a 12 segment yurt floor in plywood, and I wanted a straight edge 8 feet long to draw along, and also as a fence for a circular saw.  I happened to have some 6mm ply in 300mm widths, so I used a piece of this.

A good width for a straight edge is very useful.  It makes it much stiffer, and also offers clamping opportunities well out of the way of the business edge.


Laying out the stock on a surface

To check whether an edge is straight or not is really quite simple.  Lay the piece on a surface and draw down the edge.  I used the back of one of the plywood floor sheets. I use weights to hold down the ply, otherwise the pencil can creep underneath and give a false edge.


Draw tightly down the edge.  Use weights to avoid any gaps

Then flip it over and match it up with the drawn line.  If the edge is straight, it will perfectly match the line.  If not, the actual edge and the drawn line will be mirror image curves.  My edge was not straight, but convex.

Now I can use the drawn line to help mark a proper straight line on the stock.

I aligned the stock with the drawn line, closing the gap as far as possible.  In this case the edge was slightly convex, so the stock touched the line in the middle, leaving gaps at the ends, which I equalised.


Transferring the line

Then I set up a gauge to the widest gap.  Moving along the drawn line I transferred the drawn line at regular intervals onto the stock.  You don’t need anything fancy, a piece of wood with a 6mm notch cut out of it would do.  The main thing is to have something set up to the fixed distance. This is much better than repeated measurement.


Near the centre at the widest bulge

This left me with a line of marks on the stock that was a mirror image of the actual edge.

Then at each transferred point I marked, by eye, half way between the point and the edge.  This new line of points had to be a straight line.

IMG_0157Then it was simply a matter of joining the dots with a batten,


The straight line is half way between the edge and the transferred points

and planing down to the line.


When I retested it, the edge was spot on straight.


The drawn line fits the flipped straight edge exactly


A clearer view of the flipped line, shifted sideways

A simple but effective way of creating an accurate straight edge!

Points to remember are:

  • make sure the edge of the stock is firmly down on the drawing surface
  • use a sharp pencil for a clean line
  • use some kind of gauge to transfer the line.  A piece of wood with a cut out the thickness of the plywood would do.  The great thing is to have a fixed reference.
  • a long soled plane will be quicker and more accurate.  Clamp the stock firmly.  I just clamped it to the stack of plywood, because it was more convenient than taking it to a bench.
  • Label the straight edge!  It is easy to forget which edge you have straightened…

So having half lapped the rubbing post stock, shaped it and tidied it up a bit, it was time to bore some holes.

The posts were to be bored through the width (about 6”) with a 13 mm hole to take a length of 12mm stainless studding. The outboard edge needed to be counterbored so the nut and washer were below the surface.

Jig showing guide plate and 'legs', screwed together

Jig showing guide plate and ‘legs’, screwed together

The procedure I used for boring the holes is very familiar to boatbuilders, but with a couple of refinements.

Having set out the locations of the holes, I squared a line across the face and both edges. I found the centres of the edges, and punched the hole location on both sides. I almost always punch holes I’m going to drill. It ensures that the hole is spot on.

The principle of drilling a hole that is perfectly square to the edge and parallel to the face is to construct a guide jig that will support the drill some distance above the hole. This guide is a small piece of wood drilled to the same size at the hole. By positioning it exactly over the punched location of the hole, it will guide the drill along exactly the right path.

To construct this jig, I took a small offcut of 18mm plywood and ripped it to the thickness of the post. By sliding the post up to the circular saw blade, and the fence up the post, the exact thickness was captured on the saw. The offcut was ripped to this width.

Then I found the centreline on this guide piece, and then punched a mark in the middle on that centreline.

This I drilled out with a 13m drill.

I found two more long offcuts and screwed these to the edged of the drilled guidepiece. The jig is pretty obvious from the photographs.

The other requirement was two spacers, made from another offcut about 5 inches wide, with parallel sides, which I cut in two.

Jig initial set up

Jig initial set up

To align the jig, I lightly clamped it in place and slid the 13mm auger bit I was planning to use, through the hole, with the tip resting in the punched mark.

Spacers set up. The idea is that, because the spacer edges are parallel, the squareness of the square is 'transmitted' to the drill bit.

Spacers set up. The idea is that, because the spacer edges are parallel, the squareness of the square is ‘transmitted’ to the drill bit.

With the combination square resting on the edge, I placed the two spacers in position, one sitting in the edge and the other on the top of the jig. By pushing the jig towards the combination square (a light hammer tap maybe), the whole thing aligned perfectly.

A big gap - the jig and square are pushed together to square up the drill bit.

A big gap – the jig and square are pushed together to square up the drill bit.

A hammer can be used gently to tap it together

A hammer can be used gently to tap it together (terrible photo!)

The drill bit is now aligned to the spacer

The drill bit is now aligned to the spacer

The spacers press against the side of the auger bit and the edge of the square, the jig swivelling until everything is touching. Clamp up hard in two places.

Checking to ensure that the guide is on centre. As it was ripped to the same thickness as the stock, the straight edge should align with the the face of the stock and the edge of the guide.

Checking to ensure that the guide is on centre. As it was ripped to the same thickness as the stock, the straight edge should align with the the face of the stock and the edge of the guide.

To ensure that the guide plate was directly on centre, I set the straight edge against the face of the stock, which just touched the guide plate. If it hadn’t, I would have slackened the clamp and pushed it sideways slightly.

Rather than risk moving anything, I removed the bit from the jig to fit it to the drill, and then re-inserted it carefully. Then it was simply a matter of squeezing the trigger and letting the drill do the work.

Right through the hole...

Right through the hole…

I actually bored from both sides, first counterboring the outside with a forstner bit. But I probably didn’t have to because the holes were so accurate that you couldn’t see any join where they met.

This method is pretty good. For boring really long holes there are better ways, as any bit will tend to wander off course, diverted by the grain of the wood. A good way is to use a boring bar rather like you would bore out a cylinder on a metal lathe. This is often used for boring the stern tube hole in a boat, which has to be very accurate over a long thickness of timber.

The sides of our houseboat are vertical, and make quite a good ‘dock’ for smaller boats. I have my gaff cutter moored alongside, and also we regularly bring boats alongside. Some kind of permanent fendering seemed like a good idea.

I happen to have in my garden a large stack of keruing, which I acquired a few years ago and have been using for various projects ever since. Unfortunately the pieces, although good section at 150mm x 45mm, are only 900mm long, so are often slightly short!

Keruing is a very durable wood, often used in docks, and so I have decided to make up some longer lengths, which will be bolted edgeways and vertically to the side of the houseboat. A horizontal carpet covered board, hung from the side of boat alongside and long enough to bridge two or more of these, will ride up and down these heavy battens.

This is not precision joinery, just a means of connecting two pieces with reasonable strength. I decided on a half lap joint of 100mm. The material was left rough sawn. I cut out the waste using half depth cuts on a chop saw, 100mm from the ends, and the bandsawed to approximate depth. The idea was to plane up the joints with a router.

Some years ago I made a scarphing jig for my router. This involved a base with two rails at a 1 in 8 slope. On this rode an extended baseboard for my router. So I decided to use this baseboard to finish the half lap joints, by building a different jig with parallel rails rather than a slope.

The new jig was cut from 18mm marine ply, 12 inches wide, on the table saw. Some spare 6×2 stock was edge planed, and then ripped into two 2.5” strips. These were screwed to the plywood plywood, ensuring they were parallel.

The extended router base, showing the stiffening battens

The extended router base, showing the stiffening battens

A small piece of plastic tube prevents the router from cutting into the rails

A small piece of plastic tube prevents the router from cutting into the rails

The router base was originally made from scrap 12mm ply. The holes for machine screws to fasten to the router base were located by using the detachable router base as a template. The hole for the bit to go through the base was cut with a hole saw to the outer diameter of a piece of plastic pipe, glued in place and extending 5 mm through the base, to serve as a stop and avoid routing the rails of the base by accident! The whole thing was made more rigid by screwing edge fastened battens along it – to avoid the router’s weight causing it to sag in the middle of a cut.

When making this router base, ensure that it is long enough – it has to be surprisingly long to ensure it is well supported and stiff across the whole cut.

Squaring off the router base

Squaring off the router base

Setting the end stops from the squared router base

Setting the end stops from the squared router base

The accurate positioning of the timber to be planed is simplicity. First, the location of the shoulder of the rebate has to be set. I did this by positioning the router on the jig in a suitable location for the length of timber I was planing. Then using a square, the router base was aligned exactly at right angles to the two jig rails. Two small clamps were fastened to the rails to serve as stops. I checked that this all worked by moving the router away and then relocating it against the stops, and rechecking with a square.

Here’s the procedure to align the stock:

Aligning the router cutter

Aligning the router cutter

Set the router bit by hand so that the cutters are parallel with the rails. Lower the router to a few millimeters above the depth of cut required.

The router base is aligned to the end stops

The router base is aligned to the end stops

Set the router base against the end stops

Two plastic spacers used to align the stock parallel to the rails

Two plastic spacers used to align the stock parallel to the rails

The stock is slid into position along the plastic shims

The stock is slid into position along the plastic shims

Take a parallel edged shim and, placing it against the back rail, slide the stock along it until the rebate touches the router cutter

Take a parallel edged shim and, placing it against the back rail, slide the stock along it until the rebate touches the router cutter.  Clamp the stock in place.

The rebate in the stock touches the router cutter

The rebate in the stock touches the router cutter

If your rails are parallel, and the clamp stops are set accurately, this will locate the stock in exactly the right place to plane off the surface of the lap joint, and clean the rebate shoulder square.

Then it’s just a matter of moving the router away, dropping it to the full depth of cut and carefully planing off the remaining waste.

I had twelve of these to do. Setting up for the next cut was dead easy, as the end stops and parallel shims make it a no brainer!

Traditional rig fans w20150613_180956ill recognise this immediately.  No, not some strange Aleutian Islands carving of a marine Beast, but a jaw for our sailing dinghy Tiny Tim’s new boom.

The last boom for Tiny Tim had no gooseneck or jaw, and it always blew away from the mast on port tack.  So I thought the two hours that this project would take would be be worth it  For land lubbers, and those strange people who like bermudan rigs, the jaw is fastened by the straight edge to the boom.  The jaw and the protruding boom then form a collar that fits aft of the mast, creating a kind of universal swivel joint.  A slightly fancier vesion has a string of parrel beads around the forward curve of the mast to complete the circle.

I designed this to cut out of a piece of keruing.  Ideally the grain should be chosen to follow the curve, but I didn’t have the right shape.  I do however have loads of keruing, a tough, dense tropical harwood, so I used that and cut it on the diagonal.  For a small spar it should easily be strong enough.

The fun thing about making bits for boats is that you can to some extent make up some of the lines, and here patterns become fun things to make.

For repeat items, plywood patterns are worth making, but for one offs, a paper pattern can be quick and very useful.

In this case, the diameter of the mast is a fixed factor, as is the curvature of the boom.  But the rest is up for grabs.  On a piece of paper a compass gave me the circular space for the mast, plus quarter inch clearance all round.  The curve was drawn freehand until it felt good. So now I had my paper pattern.

The principle I have found most useful in turning an idea into a physical object is a simple one, which involves the gradual transformation of a conceptual line into a physical one.  The way this is done depends upon the tools used.  Ideally each tool rides on the surface left by the previous one.  So working out the strategy backwards is a good idea.

The finished jaw, before final shaping with abrasive paper, would be routed with round overs running on a bearing.  The router would run on a square cut ‘blank’ already satisfactorily curved.  The circle is cut with a hole saw in a drill press.  The curve will be cut roughly with a bandsaw and smoothed with a spoke shave.    The straight edge will be band sawn or cut carefully freehand with a circular saw, and then planed to the lines.

So how to transfer the lines from the paper pattern?  Easy.  Glue it on with a stick glue like Pritt.

This gluing on strategy is brilliant.  You can also use PVA glue, although Pritt dries fast and doesn’t cockle the paper much.  If you use PVA, paste the paper and let it expand before stucking it on the timber.

For a job like this, simply line up the sticky pattern with the desired grain direction and rub it on.

First I drilled the big hole on the drill press.  Then I then sawed the straight edge on the circular saw and planed to the line.  I bandsawed off the waste around the curves and tickled then up with a pair of spokeshaves.  A stationary sanding disc or clamped belt sander also does a good job of the convex curves.

The pattern is now redundant, so can be pulled off.  A light rubbing with sandpaper gets rid of most of it, the rest will come off in finishing.

I thought of routing the roundovers on the router table, but it just felt too small to work with safely, so I clamped it to a bench and used a laminate trimmer with a 3/8th roundover to get the curves on the edges.  Here I made a slight error of judgement, as I predrilled the fastening holes when it was square, thinking it would be easier and more accurate (which it was).  But of course the roundover bit bearing dipped slightly into the drilled holes.  It’s a tiny wobble in the curve, but a good reminder for the future.

Then I used 40 grit abrasive really to round out everything.  Don’t muck around with anything finer.  80 grit to rub out the marks, and 120 pre-finishing.

The straight section had to be hollowed to take the curvature of the spar.  I do have a set of hollows and rounds sitting up in my workshop loft.  But I used my handy rasp, which for such a small job was just as easy and avoided climbing a ladder.

The hole drilled in the face has a double purpose.  It serves as a fastening point for a downhaul; and whilst I am experimenting with where it fits on the boom (before permanently fastening it with copper nails and roves), I will lash it through this hole.

A coat of oil rubbed in with wet and dry, more to come.

An enjoyable little project, which I am looking forward to using.

A zen koan?

No, a common problem, and one that came up today whilst fixing a circular table.

In a world of homogenous materials, which can be cut in pretty much any way with impunity, it is easy to forget that homogenous materials are really quite modern. In the past the basic materials used for building and making things were composite and directional.  Wood is a fibrous material that changes shape irregularly with changing humidiity.  Wrought iron is also fibrous, and has to be worked with due respect for the direction of those fibres.  And stone has fracture lines and grain.

Forgetting this leads to some interesting structural failures, especially self destruction.  This post explores one such type of failure.

The mahogany circular drop leaf table in question had a split in one of the leaves. Looking at it more closely, it was clear that the glue line between two boards had opened up. Looking through the crack with the light behind revealed a dowel in the centre and two steel pins and each end.

I pulled the two pieces apart, and tried to fit them together again. There was a gap of 2 millimetres, no matter how hard I pushed.

And here’s what had happened. The dowel was holding the two pieces apart, a dowel that had clearly been inserted when the table was made.

Dowel glue failure   mortisetenon failureThis is a common problem in gluing joints. In this case, the grain of the dowel is at 90 degrees to the grain of the wood. As the wood of the table loses moisture and shrinks, the dowel stays the same length. If the end of the dowel is glued in place, the glue line in the table is literally forced open by the dowel as the table shrinks.

The same thing happens with a spline joint or a biscuit joint in solid wood. The glued, un-shrinking spline or biscuit forces the shrinking wood apart.   It’s even worse in a mortise and tenon joint. By gluing the end of the tenon in the mortise, the mortised piece shrinks back from the shoulders of the tenon and the joint becomes structurally weaker, as the shoulders no longer support it – so it will probably rack.

In each case the joint self-destructs.

Dowel successWhat is the solution? Well the explanation for the problem is the failure to understand what the glue and the spline or dowel or biscuit is for. In the case of an edge joint, like this table leaf, the glue is to glue the boards together. The dowel, spline or biscuit is to locate the boards accurately when they are being glued. Indeed the traditional way of edge joining boards is to use no spline or dowel, but a very well fitting ‘rubbed glue’ joint which doesn’t even need to be clamped up! So don’t glue the dowel at all.

mortisetenon successIn a mortise and tenon joint, the glue is to hold the joint in place. The joint should be made to be strong enough to perform its duty without any glue – indeed many mortise and tenon joints are wedged or pegged, not glued. So the glue can simply be applied to the area adjacent to the shoulder of the tenon and just inside the mortise. Then when the mortised piece shrinks, it will shrink towards to shoulder, not away from it.

Basic principle – Wood shrinks and expands a lot across the grain, but very little along it.  Join the two orientations together in a rigid relationship at your peril.