Tea House Table – Stretcher to Post

Hello and welcome to my blog!  This post continues a series in which I’m building a low chair and table for installation in a tea house.   The tea house table, a trestle table design which will reside outside, relies on wooden joinery without the use of glue.  While I have used glue in the chair build, I have decided against it for the table base.  The reason behind such a decision is a simple one, the trestle base is composed of six connections, all of them joined cross grain.  Cross grain connections stress a glue’s ability to remain elastic over time much more so than long grain connections, like those laminations used in the chair, this effect is further exaggerated by being placed outside, where connections are tested daily as the weather effects them throughout the day.

In the opening post to this series I pointed out a few changes that would be made, marking progress from the original creation of this table.  The original table featured a stretcher to post connection which consisted of a half dovetail, wedged into position.  A fellow woodworker and mentor, Chris Hall, pointed out that while this connection will function properly, its original purpose was for use in a connection where the stretcher is installed blind into a post.   The dovetail is able to be made in that scenario where access to the back of the post is blocked.

Chris pointed to a superior connection, the wedged cog joint, which provides a more assured connection.   The dovetail connection, could potentially open a slight gap when loaded, either by force on the top of the table (a lot of force!) or racking forces, such as something pushing against the top corner of the table.  Where the cog joint, with end grain abutments, will not form a gap unless the forces are so great as to compress the fibers of the upright post.  The connection is wedged in position but the wedge is not relied upon beyond simply seating the stretcher.

I began to detailed the process of cutting out the base joinery by taking short videos at a few key steps in the cutout.   The trestle table base consists of two uprights, which will be joined to the table battens and to the feet.  The base is then supported by a stretcher, connected with the aforementioned cog joints.

The cutout process begins by marking the joinery, with a marking knife and kebiki gauge, then sawing to those lines.   After sawing the shoulders I can remove the waste by paring.  I use a large usu-nomi and pare away the waste, first making a triangle shaped mound in the center, then removing that as I near the knife lines.

I have fitted the feet, but have left the joinery at the batten unfinished, doing so provides a full width reference for the marking gauges.  Moving into the stretcher to post joinery, I begin by locating the stretcher on both posts (legs) using my panel gauge.


Continuing the layout; I then work from my reference face to mark out the mortise sides with a kebiki gauge.  This gauge has locking blades, so that the width can be easily retained as I move from marking the upright post to that of the stretcher.  The stretcher is thinner than the uprights and so the gauge position must be moved.


The cog portion of this wedged cog joint is cut into the stretcher.  Marking it out is fairly straightforward, the minimum height of the cutout must be at least that of the stretcher.  However, to position the stretcher at the center of the post, the stretcher notch must be considered.  The notch allows enough room between the top of the cutout and the top of the stretcher to position a wedge into place.  I’ve used a 1/8 ratio for my wedge, as shown originally in Chris’ detail.


With all parts marked out, I can move into the cutout stage, first drilling holes at the corners of the joint and then using a coping saw to remove the majority of the waste.


Now that the sawing is completed, I can begin working the joints to their lines.   The connection between table top and uprights is a bridle joint, that bridle joint is reinforced by stub tenons which fit into the table top.  The stub tenons, a convention in Ming Dynasty Chinese furniture, work to keep the bridle joints from splaying under load and provide further leverage against forces working to the rack the legs out of position.

To form the shoulder, the outside corners are simply sawn to the line, however the material between the tenons must be chopped to the line.

After completing the bridle joinery, I complete the leg cutout by chopping out the remaining waste in the mortises.

I took a short video of an otherwise long, but enjoyable process.  The majority of the waste is chopped out, then the facets are pared into shape.

The resulting leg with joinery cut out:


I had originally intended to move the cog in from the sides, but then decided against it.  I am able to make a more accurate connection in this manner, being that I can butt the stretcher to the post, then mark the opposing side with a knife. This, opposed to measuring or paring is likely to result in a more accurate cutout and therefore a tighter joint.


Next the stretcher is prepared; the stretcher tenon was sawn to width, then pared to the lines.  This is done carefully so that the fit does not show a gap at the mortise sides.


The stretcher is then fitted into place and then marked with a knife on the outside of the post.  The section between the knife mark and shoulder are removed to make a short recess.   The recess, fitted on the post, retains the stretcher in position.


Placed in position, the resulting fit is sukima-nashi (no gaps).  Often ‘no-gaps’ is majority an aesthetic preference, in this case it is also structural.


Now that the stretcher to post connection has been formed, I can focus my attention on the connection between post, or upright, and the top.  The batten has been cut to form a bridle joint.  The post is then placed into position and knife marks are made around the tenons.  With a chisel I simply chop the waste to those lines.


The post is then placed back into position, this time with the stretcher also in place.  The resulting connection disappears, hiding its improvements over a simple bridle joint.


Refocusing on the stretcher to post joint, now with stretcher in place and the uprights positioned.  I cut out wedges for the joint and slowly fit them into position.  I am careful not to drive them in, as once that step has been taken they will not readily come out.  The top of the mortise has been carefully curved so that when it engages with the wedge, it will compress the wedge allowing it to remain in place during seasonal change.



The wedge, not yet seated, shows a gap on both sides.  To fit the wedge I simply make this gap even.  The gap will close, for the most part, when the wedge is seated.   However, since the mortise ceiling is curved it will not close entirely.


That completes the stretcher to post connection.  The next step taken in completing the table, now that the battens are permanently positioned, is to saw them back flush with the rabbet, then cut their ends at an angle.  The angled cut is made to hide the battens under the table top.


The exposed connection is to be trimmed back flush, I stopped short of flush cutting with my saw, preferring instead to trim the batten back with a paring chisel.


The finished result, shown here.  The saw marks will be left for the moment, later they will be removed when the batten end grain is detailed.


The table is then disassembled as I prepare to create square pegs which will hold the joinery in place.  The first step, after mark-out, is to drill in from both sides, creating a recess for which I can chop the mortise into a square shape.   The mortises are wedged and the batten slightly offset, so that the joint works to compress the batten and retain the post in place.


Next the pins are split out from stock, then planed into shape.


Finally I position them lightly to ensure that they fit correctly.  The bottoms shown here.


Tops shown here.  The top features two smaller pins, each below the area of the longest available short grain.


The table top is finish planed to remove any shop wear, same with many of the other parts, followed by chamfering of any exposed edges.  Finally the table can be reassembled. I placed one screw in each batten, at the center, to ensure that the wedged pins do not lift the batten out of position (dovetailed housings can only be made so tight).  Then the pins fitted to complete the batten to post assembly.  The pins are sawed and chamfered to increase their visual appeal.


Finally the wedges are seated, still the table is not yet complete.  I’ve determined that since the table will be located outside that it should be fitted with brass shoes, same as I will on the chair, to lift the table off of the ground and protect the base from sitting in water for any extended period of time.


Furthermore, I intend to detail exposed end grain surfaces with a chip carving treatment.


I’ve set aside naval brass to be used in the creation of those shoes and my next post will detail the process of machining the shoes.


Thank for following along with this build, I hope that you have enjoyed, and I look forward to reading your comments.


    1. Thanks Robert! Working on the shoes currently, it’s a fun break from woodwork. Hopefully I’ll move the Bridgeport into my shop in the future and can do a little more exquisite brass work from time to time.

  1. Great work Brian, enjoying following along. I was wondering if you could provide more details on the wedged cog joint you are using and how it differs from your original plan. Preload I’m missing it in the pictures but I couldn’t pick out quite how it works. Thanks! Keep it up


    Jonathan Elliott
    1. Thanks Jonathan! The original joint is a half dovetail it has advantages and disadvantages. The advantages being that it can be installed blind into a post and also that it does not require any extension past the post. The disadvantage is that, just like any other dovetailed connection it works to a limited degree against forces that would attempt to separate the dovetail from the socket. This effect is mitigated by the cog joint which has end grain abutments and can only be loosened by the side grain of the post either shrinking or compressing. The disadvantage is that it cannot be readily tightened after it is cut, short of placing shims or more complex method of tightening the joint.
      The top of the mortise is shaped in a slight curve to pinch the wedge in the center by compressing the side grain of the wedge as it passes through the mortise. This effect is the reason, I assume, that these connections do not loosen when used in timber structures which are certainly more prone to shrinking than furniture connections so it gives me some reassurance that the joint will remain tight.

Leave a Reply