I stumbled on to this today and thought that I would share.  Nick Engler is a life long woodworker and he does things very simply sometimes so we can understand.  He made many educational videos for Shopsmith some years ago.  

 

The test method does not seem to correspond to the typical use pattern of chair (or other wood) construction:  unclamped surfaces subjected to tension stress aren't often employed.  I think there is also confusion about cohesion and adhesion since these properties can apply differently to the glue alone versus the wood/glue assembly.  Most wood joints are sheer stress joints, where the glue locks one wood surface to another to prevent sliding parallel, and a thick (unclamped) layer of glue doesn't have the same mechanism as a thin clamped wood/glue/wood junction.  I've never seen a test where a block of glue (only glue, allowed to harden) was tested by itself (it would be an interesting challenge to form the block), but wood glues alone would fail with little strength apparent.  What makes a good wood joint is the glue's ability to raise the grain and simultaneously penetrate the grain.  In a very thin layer, the glue is simply locking the fibers of the two wood surfaces in a matrix that has good shear strength.  The moisture in the glue also causes the wood to swell, and when the glue dries, the solid remnants of the glue "freeze" the wood fibers in their swollen configuration.  The net effect is a sort of weld of fiber surfaces, and the fibers provide the strength.  In the typical M/T joint, the glue prevents the two surfaces from sliding apart (shear), and so the joint fails inside a wood surface, not at the glue.

The way I was taught to repair the chair rungs was to mix yellow glue 50/50 with water and use a syringe to inject it into the joint, around the dowel, and then a mild clamping.  The wood swelling of hole and dowel were then fixed by the very thin layer of dried glue "residue".  If the wood of the joint had worn off to the point of very loose fit, supplemental wood slivers (aka, "toothpicks") were inserted into the wet glue/wood interface, and the slivers would then bond in a dowel/glue/sliver/glue/hole matrix.  The cohesion of that matrix is pretty darn good.  At least, that chair has not come apart again.

Edited August 14, 20205 yr by PeteM
clarify

If I can remove the rung from the chair, and if the rung tenon is smaller and looser than it originally was, I'll plane a nice curled shaving from a piece of wood, and glue and wrap the shaving around the tenon, to increase the diameter of the tenon for tighter fit back into the mortise. You can either do multiple thin shavings, or one thicker one by trial and error till you get the right thick shaving. Just a nother little trick to add to this discussion.

21 hours ago, PeteM said:

The test method does not seem to correspond to the typical use pattern of chair (or other wood) construction:  unclamped surfaces subjected to tension stress aren't often employed.  I think there is also confusion about cohesion and adhesion since these properties can apply differently to the glue alone versus the wood/glue assembly.  Most wood joints are sheer stress joints, where the glue locks one wood surface to another to prevent sliding parallel, and a thick (unclamped) layer of glue doesn't have the same mechanism as a thin clamped wood/glue/wood junction.  I've never seen a test where a block of glue (only glue, allowed to harden) was tested by itself (it would be an interesting challenge to form the block), but wood glues alone would fail with little strength apparent.  What makes a good wood joint is the glue's ability to raise the grain and simultaneously penetrate the grain.  In a very thin layer, the glue is simply locking the fibers of the two wood surfaces in a matrix that has good shear strength.  The moisture in the glue also causes the wood to swell, and when the glue dries, the solid remnants of the glue "freeze" the wood fibers in their swollen configuration.  The net effect is a sort of weld of fiber surfaces, and the fibers provide the strength.  In the typical M/T joint, the glue prevents the two surfaces from sliding apart (shear), and so the joint fails inside a wood surface, not at the glue.

The way I was taught to repair the chair rungs was to mix yellow glue 50/50 with water and use a syringe to inject it into the joint, around the dowel, and then a mild clamping.  The wood swelling of hole and dowel were then fixed by the very thin layer of dried glue "residue".  If the wood of the joint had worn off to the point of very loose fit, supplemental wood slivers (aka, "toothpicks") were inserted into the wet glue/wood interface, and the slivers would then bond in a dowel/glue/sliver/glue/hole matrix.  The cohesion of that matrix is pretty darn good.  At least, that chair has not come apart again.

Well said and took most of what I was going to say.

 

Some of the major flaws in this test:

- One sample is not enough.  

- You can even see that some of the grain is parallel to the joint and some is at a diagonal by the way they split out. 

- One type of wood is not enough

- A chair joint is not long grain to long grain

- The stress test shows only one type of stress (see the attachment for different types of stress and different orientations of the wood in the joint.  Though this was on a a different joint issue, the physics are the same)

- Ignores reversibility/repeatability except for the note on hide glue.  I've heard there are two types of chairs, those with loose joints and those that will have loose joints.   Someday you may need to glue again and need to consider that

- Chair rungs are only one type of joint in a chair. 

- Chairs have a lot of joints and you need something with enough closed assembly time to get it all together and aligned before the first joints have set up.  BTDT

 

 

Leaving the ad hominem out of it, I know Nick and he is not my favorite woodworker.  

https://www.popularwoodworking.com/finishing/regluing_doweled_chairs/#

https://www.familyhandyman.com/project/fix-a-wobbly-chair-reglue-a-wooden-chair/

https://www.wwgoa.com/article/chair-repair-done-right/

https://www.motherearthnews.com/diy/home/repair-wood-furniture-dismantle-assemble-chairs-ze0z1511zhou?newsletter=1&utm_source=Sailthru&utm_medium=email&utm_campaign=12.23.15+MEN+DIY+eNews&utm_term=DIY+eNews

https://www.woodsmith.com/article/simple-chair-repair/

Biscuit Joinery.pdf

One of the tricky things in any such test is "scalability":  materials work differently at different scale.  [Modelling fluid flow is another such case:  a 1/100 scale model boat doesn't behave the same as the full size because the water isn't also 1/100, and that makes a big difference]  A thick layer of glue (which is what you get without clamping) doesn't necessarily predict the performance of a thin layer.  New concept to me:  "glue line microstructure".  mouthfull   Nice turn of phrase, tho.  I think one of the benefits of clamping is that it forces glue deeper into the grain structure, which in turn swells the wood more, and also causes greater fiber intermix between surfaces.  

     Essentially, the video shows the effect of end grain type joints, which woodworkers don't use.  (end grain joints don't interlock surface fibers, and thus you're left with "glue only" strength, much lower)

6 minutes ago, PeteM said:

I think one of the benefits of clamping is that it forces glue deeper into the grain structure, which in turn swells the wood more, and also causes greater fiber intermix between surfaces.  

You can never clamp too tightly, it's impossible, the more pressure the better, chair-makers understand this, specially the sculpted-chair making woodworkers with bent glue lamination and tight joints.

5 minutes ago, John Morris said:

You can never clamp too tightly, it's impossible, the more pressure the better, chair-makers understand this, specially the sculpted-chair making woodworkers with bent glue lamination and tight joints.

I went to what I called "Glue U." with Bob Benhke from Titebond.  He said they ran trials on red oak and said they could not "overtighten" a glue joint.  In fact, when they clamped as much as they could, it was the strongest in their tests.

2 minutes ago, kmealy said:

I went to what I called "Glue U." with Bob Benhke from Titebond.  He said they ran trials on red oak and said they could not "overtighten" a glue joint.  In fact, when they clamped as much as they could, it was the strongest in their tests.

Agreed, for my rocker glue lams I'll tighten by hand as far as I can, then I'll finish off with my channel locks, put em up on the handle and crank em again, the clamp bar is bending a tad.

The other flaw with only one type of stress test is that glue is not the same in what it can handle.  For example, CA glue is very strong in tension (pulling apart) which is why early ads showed a dude gluing his hard hat with a drop then having him picked up with a crane.  But it is weak in shear, which is why you can use it as a (removable) thread-lock substitute.

 

Another important point is "soft fail" vs. "hard fail."   If you are gluing something like a chair, you'd prefer for it to get wobbly if the glue fails and not drop you on the floor.

 

While we're talking about testing someone (I think it was Matthias Wandel) wanted to do a test using a dowel joint jig.   The vendor would only do it if it was on a specific type of wood, specific dowels, and specific glue.   They claim the superior strength of their joint, but avoid the fact that over several seasons (humidity cycles) dowel joints are known to be weak because of differential expansion/contraction.

Nick throws around adhesion and cohesion without explaining.  Cohesion is sticking to itself, adhesion is sticking to other stuff.

51 minutes ago, kmealy said:

I went to what I called "Glue U." with Bob Benhke from Titebond.  He said they ran trials on red oak and said they could not "overtighten" a glue joint.  In fact, when they clamped as much as they could, it was the strongest in their tests.

I've always been a "tighten the clamps as much as you can" guy, but some years back there ws ome hubbub about how most of us were overtightening clamps. The explanation was that such tightness squeezes too much glue out of the joint; the to-dry joint would fail prematurely. It led to some spirited discussions in some of the forums about ways to insure you didn't over tighten. I just keep squeezing the joint as much as I could. It's interesting to see the Titebond recommendation. Thanks!

Here's another guy that I've been viewing.   I don't care for all his work, especially finishing issues, but generally good stuff.  He appears to be Canadian by his slight accent and use of Lee Valley stuff.  Over different episodes, you can see the background of his cabinet behind the bench evolve.
 

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

 

 

 

Yeah, "starve the glue joint" is just another myth that won't die.