splined miter stronger than mortise and tenon

[keith ouellette]

In the feb. 2009 issue of fine wood working (I only just now read it) They did a head to head comparison of different joints by applying racking force. All samples were 3/4" thick by 2+1/2 in wide and 8 in long and glued with tight bond 3.

I was shocked to find out in their test the half lap came in first and took 1603 1b of force to break followed by a bridle joint, splined miter and then a 3/8" mortise and tenon.

The big surprise was the strength of the splined miter joint ranking #3 and the miter joint ranking #6

Another big surprise was that a simple butt joint withstood more than 150 more lb's of force than a cope+stick joint. I had alway thought a cop and stick was much stronger.

Did you ever see this article and does it make any sense to you.:confused:

[Alan Schaffter]

I also found it interesting that the Festool Domino joint was fairly low on strength.

[Pete Bradley]

Did they do more than one trial with each type of joint? I'd expect there to be a lot of variation.

[glenn bradley]

There have been a few of these tests by different mags. I believe Wood magazine did multiple failures and averaged(?). There are always some surprises (and contradictions). I could see the half lap being the big dog in a rack test due to the greater shoulder support. Other tests have different "winners"; for example:

Wood Magazine Shear test:
Mortise & Tenon Joint took 1,005 lbs/force - over double to triple the force to break compared to other joints.

Now, rack testing is different that pull-apart testing which is different than shear testing so different tests get different results. I find them all interesting and even a bit of fun. There are some videos of the older tests out there: http://www.youtube.com/watch?v=DhLfb7m9Fug

[Mark Bosse]

Two variables:

(1) Surface area of edge to edge or edge to face grain contact (ignore all end grain).

(2) Clamp pressure divided by total contact area.

Sometimes, you can have too much surface area for a given clamp pressure. Looking at 100x microscope views, I think that the glue joint thickness is related to the clamp pressure per unit of surface area of contact. Too thick a glue jont may mean weakness.

Sometimes, you can have all the pressure in the world and poor edge surface area. Try gluing end grains together and you will see what I mean. All the pressure in the world does not help.

Look at the joints in the article. There are huge differences in the amount of edge surface area. Assuming they used equal clamping pressure, I think that explains the results.

Bridle is the one that does not fit on my surface area graph. I bet you a drink that there was a very thick glue joint in that case (because they used equal pressure regardless of the surface area involved - a big mistake).

[keith ouellette]

By going back and looking at the joints and seeing where they failed from the pictures I noticed that the joint didn't fail but the wood did. Also it seemed that the very act of cutting a mortice made that area of wood weaker. The same was true of all joints where a slot or hole was cut to except the mating piece; up to a point because you also had to factor in a larger glue surface for some joints. Thats why a butt joint was stronger than a cope and stick because none of the wood was cut away to except the other piece leaving the butt joint stronger. The same logic followed with the m+t compared to the half lap. The half lap, after it was put together was more like the original solid piece of wood.

Now that I looked at it a second time It does make sense when I consider the type of force that was applied.

[Keith Christopher]

I always wonder why so many people fret over different joints and pressure before breaking. It seems they are so eager to say "this is built sturdy because I used <insert joint here> and it can stand up to..." When in the end, picking the correct joinery for the project is all that matters (not everything requires a joint to survive over 300lbs of force) . On another forum they were comparing mortise and tenon _vs_ pocket screws for faceframes. Hopefully the cabinets do not meet the force needed to require M&T joints. I think the major point is, if a chair gets racked with 1000lbs of force, it's probably going to need quite a bit of repair regardless of which joinery was used. :eek:

That being said I do find it interesting none the less.


I also found it interesting that the Festool Domino joint was fairly low on strength.

I would imagine this to be true for ANY loose M&T joint not just the domino.

[Chip Lindley]

I have built kitchen cabinet doors, using only the mating cope/stick profile to secure the joint between yellow-glued rails and stiles. A tight fitting joint with no slop holds up after years and years of use. The few that have failed were plainly abused above and beyond the call of duty.

But, would not forego mortise and loose tenons on large interior or entrance doors. Larger mass demands more support.

[Peter Quinn]

I find those test vaguely interesting and essentially useless. None of the lab tests simulate the dynamics involved with joint stress in real world conditions. At best they might catch a part of it. I wouldn't change anything I do based on anything I've seen in these tests.

Consider sitting in a chair or slamming a door. Do you apply force slowly and progressively, or do you shock and awe that chair seat and frame with the full force of your posterior? Maybe they could try letting a joint go through 8 seasons of wood movement then beating it hard repeatedly from different angles with a shock and hold force? But that's not in the budget of your typical quick and fast magazine lab test, nor is it easy to reproduce the range of real world forces wooden items face in use. Racking force? Sure, things get racked, but thats not all they get.

[Chris Friesen]

The results don't surprise me at all. The strongest joint is made when you have sufficient glue area and equalize the strength in each member (otherwise the weaker one will simply break).

The half-lap has perfectly equal strength on each side and loads of glue area. The bridle joint has even more glue area, but is trickier to equalize the strength in each member. The splined miter is basically a bridle joint using loose tenons.

The mortise and tenon joint has a few things going for it:
1) It fails gracefully when the glue fails. That is, it tends to start wobbling a bit rather than failing all at once. This is the primary advantage over half-lap, bridle, or splined joints. Once the glue fails in them, there is no mechanical join at all.
2) The blind version shows no visible joinery but is stronger than a butt joint. This is the other advantage (unless you're trying to show off your joinery).
3) It can be wedged or pinned (possibly drawbored) to avoid relying on the glue at all.

Also, as long as the joint is cut well and the glue holds, there is no reason for a loose tenon joint to be significantly weaker than an integral tenon.

[Lee Schierer]

I'm always a little skeptical of tests in magazines that use a low number of samples of any thing to draw a conclusion. Drawing conclusions from 3-4 samples is asking for trouble. To get a true picture you need to test many more samples. Wood being a natural material can vary significantly from one piece to the next with no way to tell in advance if the wood has internal flaws. I would think you would need to test 15-20 samples of a joint to draw a reasonable conclusion on relative joint strength. I'm not a mathematician, but I'm sure someone out there could add to this discussion with some statistical facts about statistics and random sampling.

[Alan Schaffter]

Yeah, we need Charlie Eppes (David Krumholz) from the TV show "Numb3rs" or Sheldon (Jim Parsons) from "The Big Bang Theory."!! to do a real analysis.

[Neal Clayton]

I find those test vaguely interesting and essentially useless. None of the lab tests simulate the dynamics involved with joint stress in real world conditions. At best they might catch a part of it. I wouldn't change anything I do based on anything I've seen in these tests.


exactly.

i can show you guys lots of examples of butt-jointed-with-a-dowel hacks that fail in 10 years or less.

they don't fail because the joint breaks. they fail because the joint traps moisture causing rot.

[Frank Drew]

Keith,

For the test, was the 3/8" tenon only 3/8" long? If so, then no wonder it failed. Stub tenons aren't a very valid test of mortise and tenon joinery, IMO.

[Doug Hobkirk]

I loved the article and was very impressed by the findings. Fine Woodworking magazine is, in my opinion, a top shelf publication and they put a lot of effort into being professional. I am skeptical by nature, so I also believe there must be large variability in wood samples, etc. Then I viewed the Dowelmax (sp?) rebuttal. They built two tenon joints and the two joints broke within 5% of each other (I was quite surprised). But I believe (fool that I may be) that some intelligent, experienced woodworkers set out to test joint strength (even if only one aspect) and were as consistent as possible constructing the various joints. I also think that FWW is the best magazine I've seen to take on such a project. Out of curiousity, and not trying to antagonize people, who would you have chosen?

I personally would have used 3 or 4 dowels, 2 Donino tenons, and probably several other changes to the joints they tested. But I suspect they used exactly what the manufacturer's instructions suggested - and I would bet those instructions would have been different if they knew those instructions were going to be used in a strength test. But that's were our common sense comes in - if the strength of a joint is very important, we would make the tenon wider, or used extra Dominos, or more dowels, etc.

I was also quite surprised that they tested as many joints as they did. E.g., pegged and wedged tenons. Almost all the joints were very strong by my standards, and I certainly learned a new respect for several new joints (and that certainly includes mitered joints).

Whew! What a rant. I get started and I just can't stop. Enjoy your next joint.

FYI - The Dowelmax rebuttal posting on SMC seems to have disappeared. For those of you who didn't see it, they felt that FWW didn't use sufficiently long dowels. In their test they used 5 dowels - they looked very beefy. I forget the number, but their joint "won" this shootout. I found the Dowelmax video quite interesting.

[David Cefai]

There was a very long thread on Knots (the FWW forum) about this same article. The really outstanding point that emerged was that in just about every case it was the WOOD that failed, not the glue.

What it therefore seems to boil down to is "How much wood is taking the strain?"

In a half-lap joint the two parts of the joint are contributing half of the wood to the joint. In an M&T one the tenon part is only contributing 1/3 of it's wood. So, given a racking force parallel to the mortised member, the M&T joint is "only" 3/5ths of the strength of the half lap.

This assumes perfectly made joints. As soon as the half-lap is not perfectly cut, smooth, glued and clamped ideally then the glue line will be weaker. Also, as was pointed out earlier in this discussion, a M&T joint will fail gracefully when the glue degrades.

I think that what this test showed was only that NEW joint strength depends on the wood area. Aging and aesthetics were not addressed.

I cannot point to the thread on Knots 'cos FWW have "improved" their website to the point of being almost unusable. One hopes that the work they are doing on it will restore it to its former glory.

[Neal Clayton]

Also, as was pointed out earlier in this discussion, a M&T joint will fail gracefully when the glue degrades.


yes, and also...

it's well established that for interior climate controlled areas on non load-bearing pieces, it honestly doesn't matter all that much. anyone who can read the forest product labs adhesive study can figure that out.

for exterior purposes, the joint design account for the shedding of water is more important than any other consideration.

therefore these tests are all in all, pretty silly.

i can glue two pieces of ply together and make them very strong and difficult to pull apart.

what does that do for me when i go to build a door or window?

a little less than nothing...

[keith ouellette]

Keith,

For the test, was the 3/8" tenon only 3/8" long? If so, then no wonder it failed. Stub tenons aren't a very valid test of mortise and tenon joinery, IMO.


There was a very long thread on Knots (the FWW forum) about this same article. The really outstanding point that emerged was that in just about every case it was the WOOD that failed, not the glue.

What it therefore seems to boil down to is "How much wood is taking the strain?"

In a half-lap joint the two parts of the joint are contributing half of the wood to the joint. In an M&T one the tenon part is only contributing 1/3 of it's wood. So, given a racking force parallel to the mortised member, the M&T joint is "only" 3/5ths of the strength of the half lap.

This assumes perfectly made joints. As soon as the half-lap is not perfectly cut, smooth, glued and clamped ideally then the glue line will be weaker. Also, as was pointed out earlier in this discussion, a M&T joint will fail gracefully when the glue degrades.

I think that what this test showed was only that NEW joint strength depends on the wood area. Aging and aesthetics were not addressed.

I cannot point to the thread on Knots 'cos FWW have "improved" their website to the point of being almost unusable. One hopes that the work they are doing on it will restore it to its former glory.


Frank.
from what I remember the tenon length was 3/8". They also had a stub tenon in the test and it didn't fair very well. And for clarification they tested 3 size tenons. 1/4" broke at 717lbs, 5/16" broke at 988 lbs and the 3/8" which was the strongest with a 1475lb break strength.


David;
Thats was kinda what I was trying to say but I didn't do it as well.