Re: Stretcher to Post Joinery
Posted: Sat Feb 27, 2016 9:22 am
As far as concerns connections in which a stretcher, 1/3rd the thickness of the post (or thereabouts), passes through the post without being shouldered or diminished in thickness, we have covered the two primary options. There is a variation which combines both dovetail and cogging approach, though it cannot be said to be a significant improvement, save for ease of mortising:
Compared to the cogged joint, the lateral registration is poorer. Compared to the half dovetail, the above version lacks any end grain abutment and has less ramp area for surface contact in one direction at least.
There is one issue worth noting with these forms of connection, in which one or two wedges are driven in over top to lock the joint: wood movement. Ideally, the stretcher would be quartersawn, however, regardless, if the sticks gain moisture, the stretcher will be trying to grow a little taller in height, however it is restricted in movement by the mortise and wedges. What will happen is that the stick will try to expand and grain in both the wedges and the edges of the stretcher will crush. Then, when humidity drops down, the joint invariably will be slightly loose.
Also, if the joint experiences racking (shear), the loading will cause forces to concentrate at the mortise openings:
This racking will crush the wedges and edges of the stretcher:
Cycles of loading, either from movement or season changes in humidity, can leave the wedges loose. This is well known in Japan. Once the wedges are loose, the mechanical integrity of the joint is obviously degraded. Obviously, if the connection is buried in a wall or generally accessible otherwise, knocking the wedges in more tightly is not always a convenient option.
One strategy that has been proposed is a threaded mechanical keeper to pre-tension the wedges and keep them at least in place:
Another strategy is to inlay metal plates on the stretcher to increase resistance to crushing:
So, these connections have been well-investigated in Japan. The use of these forms of pass-through connection is largely confined to wall bracing, where the stretchers are used often in multiples, in combination with diagonal bracing elements or not. These connection offer a convenience for putting up a structure in that they can be added in most cases after the principle framing elements are in place. They can be put in loose and un-wedged, and knocked in when the alignments of the frame are sorted out.
The cogged and the half dovetail joints also a single stretcher to pass through several posts in one piece, something a conventional shouldered mortise and tenon does not allow, and the mechanics of the joints provide good resistance to the stretcher moving laterally in either direction. A pegged mortise and tenon offers the best resistance to loading where the shoulders of the tenon take the load - tension loads, which are borne primarily by the peg alone, are much weaker.
The question about these forms of connection came about in light of an article showing a version of the dovetail being used on a sawhorse, where the stretcher simply connects two legs together. In that case, one could consider what sort of loads a structure like that must withstand, and it seems clear that the primary loading on a sawhorse is simple vertical compression. Lateral loading could be induced by placing a large timber on the horse and pushing it side to side or front to back. Of those two directions, the sawhorse is far more vulnerable to front and back loads, as these transfer directly to the joints between leg and sill pieces. Loading side to side is borne by the stretcher to post connections as well as port to top beam connections, so the sawhorse is stronger in this direction.
When it comes down to it however, if one is concerned about the sawhorse taking lots of lateral loads, in whatever direction, then it would be far better to make a sawhorse with splayed legs, so that the basic geometry of the parts transfers the load path into the ground more directly. In most cases though, sawhorses are simply bearing the weight of the loads placed atop the main crossbeam, and serve to place the timber at a convenient height for working and not much more. Given that, a standard mortise and tenon connection will suffice just fine. Indeed, if you look around Japanese shops, the sawhorses in the simple orthogonal form are invariably pegged together with mortise and tenon joints. In many cases, especially if the horse is not especially tall, the stretcher is dispensed with altogether:
There you go folks, now you see the typical shop sawhorse used in Japan. Tight joinery is vital for these sawhorses to hold together well over time, and one would not want to drag the horse around by the feet when it carries a load.
If the sawhorse needs to be laterally stiffer, a stretcher can be added as a first step. Diagonal bracing is another option (and there is a study group project from a few years back to build a braced sawhorse that you might want to take a look at). Again though, the sawhorse is actually more vulnerable to loading in the fore/aft direction, which stresses the post/foot joints. The preference really, if you want a stable sawhorse, would be to employ a splayed-post style of sawhorse. If you want a light and strong sawhorse, splayed post again is the logical choice. If you want quick and simple to build, then a beam, two posts and two feet is about all you need.
There are further stretcher connections we could look at, for cases where two stretchers or more meet the same post. I think however, in light of the original questioning about these joints in reference to that sawhorse article, the bases have been covered. If any folks are interested in looking at some of those other types of stretcher~post connections with more than one stretcher involved, I could delve into that subject a bit further. Otherwise, I hope the above set of posts have been helpful.
Compared to the cogged joint, the lateral registration is poorer. Compared to the half dovetail, the above version lacks any end grain abutment and has less ramp area for surface contact in one direction at least.
There is one issue worth noting with these forms of connection, in which one or two wedges are driven in over top to lock the joint: wood movement. Ideally, the stretcher would be quartersawn, however, regardless, if the sticks gain moisture, the stretcher will be trying to grow a little taller in height, however it is restricted in movement by the mortise and wedges. What will happen is that the stick will try to expand and grain in both the wedges and the edges of the stretcher will crush. Then, when humidity drops down, the joint invariably will be slightly loose.
Also, if the joint experiences racking (shear), the loading will cause forces to concentrate at the mortise openings:
This racking will crush the wedges and edges of the stretcher:
Cycles of loading, either from movement or season changes in humidity, can leave the wedges loose. This is well known in Japan. Once the wedges are loose, the mechanical integrity of the joint is obviously degraded. Obviously, if the connection is buried in a wall or generally accessible otherwise, knocking the wedges in more tightly is not always a convenient option.
One strategy that has been proposed is a threaded mechanical keeper to pre-tension the wedges and keep them at least in place:
Another strategy is to inlay metal plates on the stretcher to increase resistance to crushing:
So, these connections have been well-investigated in Japan. The use of these forms of pass-through connection is largely confined to wall bracing, where the stretchers are used often in multiples, in combination with diagonal bracing elements or not. These connection offer a convenience for putting up a structure in that they can be added in most cases after the principle framing elements are in place. They can be put in loose and un-wedged, and knocked in when the alignments of the frame are sorted out.
The cogged and the half dovetail joints also a single stretcher to pass through several posts in one piece, something a conventional shouldered mortise and tenon does not allow, and the mechanics of the joints provide good resistance to the stretcher moving laterally in either direction. A pegged mortise and tenon offers the best resistance to loading where the shoulders of the tenon take the load - tension loads, which are borne primarily by the peg alone, are much weaker.
The question about these forms of connection came about in light of an article showing a version of the dovetail being used on a sawhorse, where the stretcher simply connects two legs together. In that case, one could consider what sort of loads a structure like that must withstand, and it seems clear that the primary loading on a sawhorse is simple vertical compression. Lateral loading could be induced by placing a large timber on the horse and pushing it side to side or front to back. Of those two directions, the sawhorse is far more vulnerable to front and back loads, as these transfer directly to the joints between leg and sill pieces. Loading side to side is borne by the stretcher to post connections as well as port to top beam connections, so the sawhorse is stronger in this direction.
When it comes down to it however, if one is concerned about the sawhorse taking lots of lateral loads, in whatever direction, then it would be far better to make a sawhorse with splayed legs, so that the basic geometry of the parts transfers the load path into the ground more directly. In most cases though, sawhorses are simply bearing the weight of the loads placed atop the main crossbeam, and serve to place the timber at a convenient height for working and not much more. Given that, a standard mortise and tenon connection will suffice just fine. Indeed, if you look around Japanese shops, the sawhorses in the simple orthogonal form are invariably pegged together with mortise and tenon joints. In many cases, especially if the horse is not especially tall, the stretcher is dispensed with altogether:
There you go folks, now you see the typical shop sawhorse used in Japan. Tight joinery is vital for these sawhorses to hold together well over time, and one would not want to drag the horse around by the feet when it carries a load.
If the sawhorse needs to be laterally stiffer, a stretcher can be added as a first step. Diagonal bracing is another option (and there is a study group project from a few years back to build a braced sawhorse that you might want to take a look at). Again though, the sawhorse is actually more vulnerable to loading in the fore/aft direction, which stresses the post/foot joints. The preference really, if you want a stable sawhorse, would be to employ a splayed-post style of sawhorse. If you want a light and strong sawhorse, splayed post again is the logical choice. If you want quick and simple to build, then a beam, two posts and two feet is about all you need.
There are further stretcher connections we could look at, for cases where two stretchers or more meet the same post. I think however, in light of the original questioning about these joints in reference to that sawhorse article, the bases have been covered. If any folks are interested in looking at some of those other types of stretcher~post connections with more than one stretcher involved, I could delve into that subject a bit further. Otherwise, I hope the above set of posts have been helpful.