Stretcher to Post Joinery

What's on your mind?
User avatar
Chris Hall
Site Admin
Contact:
Location: Greenfield, Massachusetts
Posts: 2385
Joined: Mon Nov 22, 2010 12:46 pm

Re: Stretcher to Post Joinery

Wed Feb 03, 2016 11:06 am

In the preceding examples, we looked at the half dovetail connection, done blind. As noted there, a longer dovetail and deeper mortise mean greater surface area for interlock and thus the longer example should be the stronger example.

Taking this idea a step further, we could bring the half dovetail right through the receiving member:
Nuki to post connections through.jpg
Nuki to post connections through.jpg (187.07 KiB) Viewed 7416 times
Note that the wedge hasn't been detailed in the above drawing, and there is no mortise provided for the wedge, so the above is simply for sake of example.

To maximize the dovetail, it is extended to the exit face of the receiving member. It's hardly against the law or anything like that, but there is no reason to configure the connection so that the dovetail ramp surface sticks out any further. With a portion of the stretcher protruding beyond the receiving piece, it is more desirable to present the stretcher as a clean pair of horizontal lines passing through, and not have the joint mechanism be any more obvious than it already is. remember, this form of connection is generally employed on framing that is not visible to view, and so the aesthetics are less important in that context. If you choose to adapt it for furniture work, there is absolutely nothing wrong with that, however there are plenty of alternatives. The loading on the parts should guide the joinery decisions ahead of aesthetics in most cases, and there may be several options which meet performance goals and yet offer quite different aesthetics.

With the connection passing fully through the receiving member, we have options to wedge the connection as before, from the inside face:
Nuki to post connections through 2.jpg
Nuki to post connections through 2.jpg (321.92 KiB) Viewed 7416 times
Or, we can drive the wedge in from the outside face:
Nuki to post connections through 3.jpg
Nuki to post connections through 3.jpg (293.89 KiB) Viewed 7416 times
Notice that in either case, if we keep the wedge taper the same, at 1:8, the lengthening of the wedge can result in one end of the wedge becoming quite large. This may be objected to on aesthetic grounds in cases where the connection is exposed and not buried in a wall, but also there is the issue that, as the wedge section becomes larger, it becomes prone to greater overall seasonal movement. We could make the wedge ramp angle shallower, say 1:10 or less, to bring the dimension of the fat end down a bit, however the trade off there is that the friction between the wedge and abutment goes up, making it a little harder to fully seat the wedge and definitely more difficult to remove the wedge should the need arise. Also, if the wedge requires more force to drive in, then the potential for damaging the end of the wedge by insertion increases. These are not huge issues, and are less a factor with softwoods than hardwoods.

Another issue concerning aesthetics relates to the end of the wedge relative to the end of the stretcher. Should they be flush? Probably not, as you would likely want to chamfer the arrises around the end of each piece. Having the wedge project beyond the end of the stretcher doesn't make much sense, but if it is to be held back from the stretcher end, then by how much? There is no set answer to that. Generally, the wedge looks best, I think, if it is proud of the post the same amount on entry and exit face, so one might think simply in terms of having the stretcher extend a bit further than the end of the wedge, and that is all.

Otherwise, whether one drives the wedge in from the outside face or inside face is largely a matter of which face provides more convenient access for the hammering.

Another issue arises with driving a wedge though this sort of connection. Here is a video which shows the problem clearly:

https://www.youtube.com/watch?v=rybMmT10rhw#t=52.819174

Notice how the wedge starts to blow out a piece of wood on the exit face? In the next post we will look at that situation a little more and how it might be dealt with, and then discuss other wedging options.
User avatar
Chris Hall
Site Admin
Contact:
Location: Greenfield, Massachusetts
Posts: 2385
Joined: Mon Nov 22, 2010 12:46 pm

Re: Stretcher to Post Joinery

Thu Feb 04, 2016 11:43 pm

In the previous post we looked at a couple of options for a through version of the half dovetail:
Nuki to post connections compare half dovetails.jpg
Nuki to post connections compare half dovetails.jpg (511.4 KiB) Viewed 7399 times
In the interest of being thorough, there is more to explore with this connection yet.

Note that the wedges used in both joint version pictured are exactly the same size and slope, and sit proud of the post by the same amount on each side.

Now we'll take the wedges out:
Nuki to post connections compare half dovetails 2.jpg
Nuki to post connections compare half dovetails 2.jpg (440.22 KiB) Viewed 7399 times
Next, let's withdraw the stretchers up the mortise ramp and see what happens, remembering from an earlier illustration that we need the mortise to allow the entry of the stretcher:
Nuki to post connections compare half dovetails 3.jpg
Nuki to post connections compare half dovetails 3.jpg (419.92 KiB) Viewed 7399 times
You can see clearly that the form of this joint in which the wedge is driven in the same direction in which the stretcher enters the post ( the background assembly) has no points of interference at the mortise opening when the stretcher is drawn back to present its maximum height. Not so with the version in the foreground, in which the wedge enters from the exit face of the stretcher - note the arrow showing where the mortise interferes with the stretcher. The mortise therefore would have to be made taller to accommodate the same stretcher, or the stretcher shrunk down in height. Since a stretcher is a structural component, we will assume the height is the height it needs to be for strength, and so therefore the only option really is to make the mortise taller, all things being equal between the two joints.

The upshot then is that driving the wedge from the backside of the post means that the mortise must be taller and the wedge therefore is also taller to fit the opening.

In this example, the mortise on the foreground example has been adjusted upwards, along with the wedge, which has been thickened:
Nuki to post connections compare half dovetails 4.jpg
Nuki to post connections compare half dovetails 4.jpg (439.05 KiB) Viewed 7399 times
And removing all the parts so we can compare just the wedges:
Nuki to post connections wedges compared.jpg
Nuki to post connections wedges compared.jpg (120.85 KiB) Viewed 7399 times
So, this demonstrates fairly clearly, that if you have the option to insert the wedge from the same side as the stretcher enters from, you will gain the benefits of a smaller mortise and slimmer wedge. The smaller mortise makes for a stronger post, and the slimmer wedge is going to be less affected by seasonal movement as would the taller wedge. The advantage is clear in that respect.

One point in favor of having the wedge enter from the exit face is that any loading on the stretcher which exerts tension - tries to withdraw the stretcher - will tend to tighten the wedge automatically. It has a slight mechanical advantage in that sense.
User avatar
Chris Hall
Site Admin
Contact:
Location: Greenfield, Massachusetts
Posts: 2385
Joined: Mon Nov 22, 2010 12:46 pm

Re: Stretcher to Post Joinery

Fri Feb 05, 2016 10:32 am

A couple of posts back I made note of the issue of the chance for blow out on the exit face when the wedge is driven in. Obviously, one could clamp some sort of wooden block in place to help keep things together.

Another strategy is to change the shape of the mortise abutment for the wedge so as to place a crimp point before the exit face:
Nuki to post connections bowed abutment.jpg
Nuki to post connections bowed abutment.jpg (406.74 KiB) Viewed 7387 times
Look closely and you can see that the abutment has been made to a slight downward curvature. It doesn't have to be a curved surface however, it could also be a facet.

By moving the pressure point more to the middle, instead of trying to make it a flat abutment, we are adopting a similar idea to that of the chip breaker and fixing pin on a kanna: place the pressure point in the middle so as to push down evenly and so that there are no other points where things could get hung up. The degree of curve or bulge in the abutment would be a matter for investigation, but it need not be too pronounced - perhaps 1~2mm at the most.

Here's a Japanese graphic I came across showing the same thing:
nuki detailing-small.JPG
nuki detailing-small.JPG (410.86 KiB) Viewed 7387 times
They seem to have goofed up on the tenon orientations for the post there, but that is besides the point.

That covers the half dovetail form of connection. We'll next look at the wedged and cogged joint between stretcher and post.
Marc
Raw Log Import
Posts: 24
Joined: Sat Nov 29, 2014 8:35 am

Re: Stretcher to Post Joinery

Fri Feb 05, 2016 11:10 am

Great infos, thanks.

One question about the endcuts of the wedge - is there any particular reason why one end is cut perpendicular to the sloped surface and the other end perpendicular to the straight surface of the wedge?
Which leads me to another question - the grain orientation of the wedge.
Should the grain run parallel to the sloped surface or parallel to the stretcher?

Hope these questions are not too silly. :oops:

Looking forward to the next explanations.
User avatar
Chris Hall
Site Admin
Contact:
Location: Greenfield, Massachusetts
Posts: 2385
Joined: Mon Nov 22, 2010 12:46 pm

Re: Stretcher to Post Joinery

Fri Feb 05, 2016 11:11 am

渡り顎 Watari Ago.

Watari means "cross (over)", and ago means "jaw".

Like the connection described previously with the half-dovetail, this joint can be used at the termination of a stretcher or as a mid-point connection.

The parts:
Nuki to post connections wedged cogged a.jpg
Nuki to post connections wedged cogged a.jpg (253.94 KiB) Viewed 7384 times
The stretcher slides in laterally:
Nuki to post connections wedged cogged b.jpg
Nuki to post connections wedged cogged b.jpg (244.92 KiB) Viewed 7384 times
Then the stretcher drops down over the cogged portion:
Nuki to post connections wedged cogged c.jpg
Nuki to post connections wedged cogged c.jpg (249.75 KiB) Viewed 7384 times
And then the wedge secures the connection - again, the joint may be detailed such that the wedge may be inserted from entry or exit face:
Nuki to post connections wedged cogged d.jpg
Nuki to post connections wedged cogged d.jpg (277.24 KiB) Viewed 7381 times
An interior view makes the relationship between the parts clear:
Nuki to post connections wedged cogged e.jpg
Nuki to post connections wedged cogged e.jpg (347.51 KiB) Viewed 7384 times
The mortise opening height must, at a minimum, be equal to the stretcher section height:
Nuki to post connections mortise opening.jpg
Nuki to post connections mortise opening.jpg (372.03 KiB) Viewed 7371 times
Probably best to allow an extra mm or two so as to make the fitting a little less difficult.

To control lateral movement of the stretcher, or, put another way, bending in the post, the half dovetail approach offers a 90˚ abutment in one direction, and a 1:8 or so dovetail angle in the other direction. The 'crossing jaw' joint has a 90˚ abutment in both directions, which provides a more positive registration of the stretcher to the post.

When the half dovetail connection has tension forces acting against the dovetail, the wedging action will cause the surfaces involved to slide and deform. As the post's abutment against the stretcher's dovetail is all end grain, the crushing accrues almost entirely with the stretcher dovetail. When the tension load is released, the dovetail will uncompress a slight amount, but with remain compressed as compared to previously. So the joint will be loose in both horizontal and vertical axes.

With a cogged joint, when strong tension or compression loads are applied to the stretcher, the stretcher bears with an end grain abutment against small portions of the post's long grain, which are softer. There is almost no slipping of the surfaces by one another. There is a limit to how much crushing can occur, and almost all of that will happen with the post's lateral joint abutments - - I would suggest the total amount that a cogged joint will let the stretcher withdraw is going to be less than the half dovetail joint will allow it to move.

The half dovetail has an advantage as a terminal connection for a stretcher where there is no room on the backside of the post for any part of the stretcher to protrude. The dovetail can be done a bit more than halfway in, 3/4" of the way in, or through and cut off flush. The cogged joint does not have this versatility, and must be provided with an adequate amount of relais beyond the cog notch or the end of the joint would pop off (short grain) if strong tension loads were imposed on the stretcher.

Next, we'll look at more wedging configuration options - -these can apply to the last couple of joints described above, half dovetail and cogged.
User avatar
Chris Hall
Site Admin
Contact:
Location: Greenfield, Massachusetts
Posts: 2385
Joined: Mon Nov 22, 2010 12:46 pm

Re: Stretcher to Post Joinery

Fri Feb 05, 2016 11:57 am

Marc wrote:One question about the end cuts of the wedge - is there any particular reason why one end is cut perpendicular to the sloped surface and the other end perpendicular to the straight surface of the wedge?
Which leads me to another question - the grain orientation of the wedge.
Should the grain run parallel to the sloped surface or parallel to the stretcher?
I am grateful for the questions.

The end cuts of the wedge, what angle and why? The slim end is simply cross cut 90˚ relative to the bottom of the wedge. The cut end would be chamfered on its three upper surfaces.

The big end of the wedge is different, as it is cross cut 90˚ to the wedge's sloped surface.

There are a couple of reasons why the end cuts are done this way.

One is that it reduces the chance of splitting off a piece of the upper portion of the wedge by hammering upon that corner at the wrong angle.

If the fat end of the wedge was oriented 90˚ to the bottom surface of the wedge (i.e., the same way the slim end of the wedge is cut), one would need to swing the hammer directly in line and upon the stretcher's upper face, at a perfectly horizontal orientation, to drive the wedge in. And if the hammer maybe bumped down and then sprung up slightly off the surface it could end up traveling slightly upward and be striking the upper corner of the wedge a more glancing type of blow. The corner of the wedge could break off due to the short grain support, and the wedge will be more prone to developing splits after repeated hammer blows.

This splitting or flaking off of the wedge could even happen by accident if you were using a wooden drift to set the wedge.

Another is that sliding the hammer along the edge of the stretcher can mar the face of the stretcher. You saw in the previously-linked video the worker bash the stretcher (and even the post face on one occasion) several times as he set the wedge. One could place a wooden piece on top of the stretcher to protect it's surface from the hammer, but that of course reduces the space available to hit on the end of the wedge. Better to not orient the end cut of the wedge at 90˚ to the bottom surface of the wedge.

One wants to swing the hammer in a path which is more likely to avoid contacting the stretcher edge, or at least a smaller portion of it, by swinging in an angle some amount above the horizontal. Orienting the end of the wedge so that it is perpendicular to that angled swing means a more even reception of the the hammer's energy and less chance of messing up stuff around it.
Should the grain run parallel to the sloped surface or parallel to the stretcher?
The grain of the wedge is ideally 90˚ to the hammer face striking angle and in line with the mortise ramp. Parallel to the sloped surface.

If the grain of the wedge ran perfectly parallel to the surface of the stretcher, it would be a minor matter however, as the maximum slope of grain would be at 1:8 or thereabouts, which is fairly modest.

Here's a video showing a guy setting some wedges - he almost never mars either stretcher or post face:

https://www.youtube.com/watch?v=6_NkIOLlmrc

Those are kinda fat wedges for my taste, but making them fat presents a larger target to hit.

Another video shows what looks to be a wedged half dovetail, with the dovetail end flush with the post:

https://www.youtube.com/watch?v=vSHQls9xioI

He's putting together a connection there which has opposed wedges used - we'll talk about that soon enough.
Marc
Raw Log Import
Posts: 24
Joined: Sat Nov 29, 2014 8:35 am

Re: Stretcher to Post Joinery

Fri Feb 05, 2016 12:27 pm

Thanks for the detailed answer.
That makes sense.
User avatar
Jon B
Sweeper of Floors, Maker of Tea
Location: Brooklyn, NY
Posts: 69
Joined: Tue Nov 25, 2014 7:33 am

Re: Stretcher to Post Joinery

Fri Feb 05, 2016 9:57 pm

Thanks for the excellent details on these joints Chris, looking forward to the continuation of this series. Is there a rule of thumb for how much relish to leave on a stretcher notched for a cogged joint? In your drawing it looks like you left at least as much material after the notch as was cut out.
User avatar
Chris Hall
Site Admin
Contact:
Location: Greenfield, Massachusetts
Posts: 2385
Joined: Mon Nov 22, 2010 12:46 pm

Re: Stretcher to Post Joinery

Fri Feb 05, 2016 11:15 pm

Jon B wrote: Is there a rule of thumb for how much relish to leave on a stretcher notched for a cogged joint? In your drawing it looks like you left at least as much material after the notch as was cut out.

I think that depends upon the application, just as it does for a pegged mortise and tenon joint. For comparison's sake, a m&t which deals with compression loads typically might have 3x peg diameter for relais, but if the joint is to endure tension loads, then 7x peg diameter is a more appropriate number. It's hard to quantify the amount of relish directly for the watari ago joint , and I haven't come across any scientific testing on the matter or hard data, so I can't give more than a general answer. The more tension the joint is exposed to, the longer the relish the better. The failure mode -shear failure parallel to grain - is clear to see however.

There is quite a lot in the way of of Japanese scientific experiment with the performance of stretcher joint connections at posts and their performance under seismic loading conditions, by the way.
User avatar
Chris Hall
Site Admin
Contact:
Location: Greenfield, Massachusetts
Posts: 2385
Joined: Mon Nov 22, 2010 12:46 pm

Re: Stretcher to Post Joinery

Wed Feb 10, 2016 9:59 pm

More to come by week's end...

Return to “General Discussion”

Who is online

Users browsing this forum: No registered users and 3 guests