Dowel Joinery?

[Thomas McCurnin]

I haven’t used dowels or biscuits in years, although I still have my 40 year old doweling jig and my Porter Cable biscuit machine. I find that properly jointed edge joints with Titebond works good enough for me. While in theory they help with alignment, I sometimes find the opposite is true—if the hole or biscuit mortise is not 100% accurate, I’ll get lippage which can’t be fixed by crawls and clamps. At least with a naked Titebond joint what you see is what you get.

For chairs, I wouldn’t use anything but a mortise and tenon.

I did have an occasion to repair some cheap 30 year old furniture and found that the dowels had shrunk and failed. I plugged the holes with new wood and epoxy. Then re-drilled the dowel holes, found perfect sized dowel stock (not easy) and epoxied those in place. It’s holding under high weight.

I would not voluntarily use biscuits or dowels unless the design or glue up required it.

[Mike Henderson]

Thanks for the math. Sounds good in theory.

Do you have a comment on the source articles from Flexner and Wandel?
Wandel's test in particular reveals a real world outcome very different from what your math implies (in theory).

Ironically I repaired a chair about three weeks ago that the customer bought from Restoration Hardware. The mortise and tenon joint failed. There appeared to me to be two problems. One was that the mortise and tenon joint had left so little "meat" on either side of the mortise that it broke right through the back leg. The other looked to be what might have been insufficient glue.

Again, not lobbying for one kind of joinery over another 100% of the time.
One thing people don't always consider with a mortise and tenon joint is the wood that's being removed from one member and whether there is sufficient surrounding wood left to handle whatever load is transferring through the tenon. When I've seen mortise and tenon failures it's been for this reason primarily. I think both types of joinery have their place. The key is to use each one where appropriate.

One other point, I think two dowels, especially if they were not coated completely with glue, may be culprit behind your repairs more than the whole method of joinery being unsound. One thing I like about the Jessem system is the ability to do rows of dowels, in an offset pattern to minimize surrounding wood failure. If the chairs you have been repairing had been joined with a row of three or four dowels and both the hole and dowel were brushed with glue like you or I would do in our shops, I think you would see fewer failures. Again, I'd reference you to Flexner's article who does a better job of explaining it.

Edwin

More than theory - I've seen this in practice.

I have not seen ANY chairs in for repairs that showed signs of not having glue on the whole dowel. And I would not expect to see that. The companies who manufacture chairs are not incompetent. As I said several times before, what I've encountered is broken wood, not glue failure. When I've taken the joint apart there is wood still stuck to the glue of the dowel. The failure is wood failure, not glue failure, and the problem is insufficient glue surface.

In my calculations, I've assumed 3/8 inch dowels and tenons that are 3/8 inch thick. I have not experienced any failures where the wood of the chair side or back broke away. The stress on that chair joint is generally levered apart (from people rocking back), and not side-to-side.

My point is that a mortise and tenon that fits into the same space as two (or more) dowels is significantly stronger than the dowels. Those woodworkers who want to create long lasting chairs would be advised to use M&T and not dowels.

Mike

[And regarding chair joint failure - Thank Goodness for corner blocks. Sometimes that's the only thing holding the chair together.]

[Mel Fulks]

I have no doubt that Mike's figures are correct. When it is deemed best to use dowels to replace a broken tenon, the
dowels need to have larger diameter. One cause of failures ,and I mean looseness as well as complete failure, is an ill
fitting of the two meeting surfaces.

[Mike Cutler]

I've repaired a number of antique doors through the years. Just got done with one over 100 years old last year. The dowels are big, 1/2" diameter and about 8" long. They go deep into the rails and stiles of a door,and are usually spaced about 2"-3" apart. I've never repaired a broken dowel, or had them come out easily. They almost always broke/split the wood on the face of the stile.
Dowels are like any other joinery technique. They have their applications. They do require a significant effort to maintain symmetry and uniformity of a joint, or it won't go together. Antique doweling machines, when you can find them, are still very expensive.
They will not equal the strength of a properly sized and fitted M&T joint, but they will be stronger than the two pieces of material they are joining together, if sized properly.

[Mike Henderson]

Another thing that supports the glue surface area problem is that when I repair a chair with two dowels at the back joint, the dowels that are into the seat side are almost always still solidly glued into the side. The side that has come loose is the dowels into the chair back.

The reason is that the dowel into the side of the chair is all long-grain-to-long-grain surface area because the dowel goes into the wood along the grain, not across it. The dowels into the back of the chair go into the side of the wood, across the grain, so half of the dowel is facing end grain. The dowel into the side of the chair has twice the long-grain-to-long-grain surface area compared to the dowels into the back of the chair.

Mike

[Von Bickley]

I'm just a old retired disabled woodworker and cannot justify buying a Domino, so I will continue using my dowel jig. I have used dowels on several projects and have no problems with alignment or strength.jessem dowel jig.jpg

[tom lucas]

I've used dowels for years on edge to edge joinery. Good as any other, even if slow to make. I've not had a failure yet. However, I only use them in end-to-face joinery when nothing else would work. I'm with Von on the domino thing. I cannot justify the cost of a domino. I'll use dowels, biscuits, or make mortise & tenon joints. Biscuits are really fast, but not as strong as dowels. So, I mainly use them where there a good number of joints that don't see a lot of stresses. I do get how the domino is useful, and likely worth every penny for professionals where time is money.

[Will Blick]

Mike, interesting take on the dowels gluing to end grain over half their surface area. I am curious if you ever viewed the videos and explanations on the DowelMax site showing how Dowels are stronger than MnT. thoughts?

While I certainly see the inherient strength of a MnT joint, I am curious if you feel the multi dowel joint would have its own set of unique benefts. A MnT joint is a single joint. When the two mating surfaces expand and contract at slightly different rates over time, this runs the risk of the glue bond breaking or weakening the ENTIRE joint. I have witnessed this in old furniture. Specially when u join two different wood species. With dowels, the surface areas is so small for each individual dowel, (compared to the single MnT length) it seems dowels are much less vulnerable to this type of bonding failure.

I would assume in a majority of applications, the breaking force of either MnT or dowels FAR exceeds any real world forces that would be applied to it. So proving one or the other will break at 3k lbs vs. 4k lbs, is a moot point when the joint may only be subjected to 200 lbs max in the real world. Therefore, possibly maintaining the glue bond over time is the more critical factor. would u agree?

On the issue of panel glue ups,
I often like to clamp tight, to assure seamless glue joints. Its so rare to get PERFECTLY straight edges...and even when you do, within a few hours, the wood moves. This is why TiteBond recomends 150 to 250 psi on the joint lines. Its to overcome non straight edges, not for the glue to bond. This is a LOT of force, which is a lot of clamps with 3/4 material.

When I clamp with NO joinery, the boards can slip n slide a bit, its inevitable. I have tried end clamps, massive allum. bars across the glue lines, with there own set of clamps, but the wood can still creep, requiring more work to flatten the surface of boards that started out perfectly flat.

I agree with a previous poster, biscuits are not ideal for this, despite their long standing repuation for this task. Sometimes, biscuits work, other times, they cause more problems than benefits, specially with fast drying glue and inconsistent thickness of the biscuits. Loose tenons are a good option for sure as you only need a snug fit in one direction, so it is relatively fast, on par with biscuits. I have had good luck with the Domino for this task, albeit, you consume a lot of $ in dominos for panel glue ups. Dowels are dirt cheap and fit really snug. And with the dowelmax, its not that difficult to align the holes perfectly on both boards. With a dry fit, you avoid any alignment surprises during glue up. I rather put the time into the glue up, vs. the clean up of the glued panel.

[Gordon Stump]

I use dowels when I have to. Like on the large mitered frame of a gaming table I recently made. Or the arched door of Shaker tall clock I made years ago. Without a biscuit joiner or loose tenon machine, using dowels made sense. Not my first choice but plenty strong. My biggest problem with them is the time it takes to glue up. Recently I used 24 hour T-88 and that solved that.

[Mike Henderson]

Mike, interesting take on the dowels gluing to end grain over half their surface area. I am curious if you ever viewed the videos and explanations on the DowelMax site showing how Dowels are stronger than MnT. thoughts?

DowelMax would hardly be an unbiased source for the comparison of dowels to M&T, now would they?

Dowels have their place and have been used for many years. A dowel joint can be sufficiently strong for a particular application. The key is to make sure you have sufficient long-grain-to-long-grain glue surface area for the particular joint and the stress on that joint.

My comments were specifically about chairs and the way the seat is joined to the back with two 3/8 inch dowels on most commercially made chairs - it's simply not strong enough and the reason is because there's just not enough long-grain-to-long-grain glue surface area. If they put in more dowels, perhaps in a zigzag pattern, they could create a strong enough joint. I suspect they don't do it for cost reasons - and because the dowel joint lasts about as long as most people keep chairs.

But if the chair manufacturers wanted to create a stronger joint, a loose tenon would probably be quicker and easier to manufacture than putting in more dowels.

My advice to anyone making a chair is to use a M&T joint in the back of the seat, instead of two dowels, if they want the chair to last.

Mike

[For lining up boards in a panel glue up, consider the use of cauls. They do a really good job of keeping the boards aligned during a glue up. Another approach (a bit slower) is to do two boards at a time and then add boards one-at-a-time to the first two boards. That allows you to do a better job of keeping the boards aligned.

None of those alignment techniques - dowels, biscuits, Dominos, etc. - add any strength to the joint and they take time to do. Also, you can run into a problem a friend of mine had. He used biscuits and wound up having to trim the panel in such a way that he cut through a biscuit (it showed on the edge).]

[Edwin Santos]

More than theory - I've seen this in practice.

I have not seen ANY chairs in for repairs that showed signs of not having glue on the whole dowel. And I would not expect to see that. The companies who manufacture chairs are not incompetent.

I don't think anyone is saying the companies that manufacture chairs are incompetent. What Bob Flexner (certainly not incompetent) is saying in the article I linked is that manufacturers are working around efficiency constraints that are sometimes defined by their machinery and cost pressure due to market competition. The operating theory they are using is to drill a hole with one machine, and squirt a glob of glue into it with another machine, and insert a fluted dowel with yet another machine. The expectation is that the glue will wick up the fluted dowel, which happens to some degree and is why you are seeing some glue on the whole dowel. However this method is considerably different than the care you or I would take in our own shops where we might use a brush and liberally coat the dowel and the hole with glue.

This is why I liked Matthias Wandel's test because being a one man small shop, I'm sure he was applying glue to all joints just as we would do, not using the factory methods. And of course his outcome was that the strength of a dowel joint and a mortise and tenon joint were surprisingly similar.

Anyway, I am not promoting a dowel vs. mortise and tenon argument, I am a fan of dowels where they are appropriate and I will say I think they are much stronger than you suggest. If they were not, then James Krenov and his legions of Krenov school alumni are missing the mark, as is Brian Boggs who builds his chairs exclusively with dowels. And Sam Maloof used an elaborate joint held together with screws which could only rate poorly using your math as a criteria.

Edwin

[Will Blick]

Mike, after 30+ years in the business world, I am wise enough to realize who is putting out test results and the chance of bias. However, just because a partial party shows a test result, doesn't in itself, make the test a sham. I have NO dog in this fight... as I mentioned, I use everything...the right joinery for the job. I was just curious if you have seen the test, and if you had any comments on it. I was complimentary about your methodical approach towards this issue in my post. I guess, my bad for asking. I will refrain in the future.

I marvel that a few hot topics with ww's, such as dowels vs. dominos, vs. biscuits, or, number of clamps for panel glue ups....
these topics seem to force some people to bunker down in their camps, and defend at all costs. I will never understand this, but such is the world of forums! Not just ww, Ford vs Chevy, Apple v Android, PC v Mac, etc.

[Mike Henderson]

And Sam Maloof used an elaborate joint held together with screws which could only rate poorly using your math as a criteria.

Edwin

I can't possibly see how you got to this from what I posted. Screws are entirely different from dowels and M&T and joints using them can be quite strong.

I have repaired a lot of doweled chairs and I've observed how the dowels failed. As I've said many times before, the failure mode is not glue failure (or missing glue) but wood failure. There is wood stuck to the glue on the dowels which indicates that the dowel has been pulled out of it's hole. And since the glue is stronger than the wood, the wood broke when the dowel was pulled out. And there was glue on the whole dowel - solid dowels, not those with spiral or fluted grooves. It is very clear that the failure is not due to poor assembly or poor glue application.

So I started analyzing why that happened, which led me to realize that there just wasn't enough glued surface area to withstand the tension forces. Then I analyzed the surface areas and realized that only half of the dowel is getting glued to long grain.

What other people do when they build furniture is up to them. For me, it's easy to see that two dowels on the back of a chair seat is not strong enough and that a M&T in place of that will be stronger. The mathematics I provided will tell you how much stronger.

Mike

[Let me go through some analysis.

First, just about everyone has encountered dining room chairs where the back joint has failed. A simple test is to put your knee on the seat that pull and push the back from the top. If the back moves back and forth, the joint is almost always loose.

Second, we see this in dining room chairs of all different designs (but much more often in chairs without stretchers), all different woods, and all different manufacturers.

Third, when we disassemble the chair we usually find that the joint consists of two 3/8 inch dowels, and the dowels are loose in their holes. Most of the time, the loose part is in the back, not in the side.

Removing the dowels, we find that they are different wood species and that they show evidence of glue on the whole surface. They are not fluted or spiral dowels. Further, we find pieces of wood still attached to the glue on the dowels.

What is our theory of the failure?

Your theory is that the glue was not applied properly to the dowels or the holes. Let's examine that. To accept that we'd have to have many manufacturers in many different countries all with the same problem of improperly applying glue to the dowels and dowel holes. Further, we should have reports that the dowels removed from those failed chairs are dry, at least on a significant portion of the dowel. However, I have not heard any such reports and have not seen that myself. Additionally, I believe that chair manufacturers are more competent than that. I can think of several ways of applying glue evenly to the dowels and holes with a machine, and controlling the amount of glue for cost reasons and to avoid glue escaping and being visible after assembly. It's very hard to believe that they would be so crude as to just shoot a wad of glue into the dowel hole. I would rate that theory as unlikely.

Another theory is that the manufacturers do a good job of applying the glue but there is just not sufficient glue area on two dowels, especially on the chair back, and the dowels are being pulled out, breaking the wood that the glue is attached to. In support of that we can calculate the amount of long-grain-to-long-grain glue surface area and compare it to a M&T joint. From observation, most of the time the dowel is pulled out from the chair back, and not the chair side which has more glue surface area. Additionally I have reported that most of the dowels removed from failed joints have wood still stuck to the glue of the dowel.

If you wish to continue with your theory that the chair manufacturers are not glueing the dowels properly, you need to provide some evidence other than hearsay.]

[Edwin Santos]

I can't possibly see how you got to this from what I posted. Screws are entirely different from dowels and M&T and joints using them can be quite strong.

I have repaired a lot of doweled chairs and I've observed how the dowels failed. As I've said many times before, the failure mode is not glue failure (or missing glue) but wood failure. There is wood stuck to the glue on the dowels which indicates that the dowel has been pulled out of it's hole. And since the glue is stronger than the wood, the wood broke when the dowel was pulled out. And there was glue on the whole dowel - solid dowels, not those with spiral or fluted grooves. It is very clear that the failure is not due to poor assembly or poor glue application.

So I started analyzing why that happened, which led me to realize that there just wasn't enough glued surface area to withstand the tension forces. Then I analyzed the surface areas and realized that only half of the dowel is getting glued to long grain.

What other people do when they build furniture is up to them. For me, it's easy to see that two dowels on the back of a chair seat is not strong enough and that a M&T in place of that will be stronger. The mathematics I provided will tell you how much stronger.

Mike

I think at this point we can just agree to disagree and build according to our own method preferences.

Edwin

[Patrick Walsh]

Figured I’d add to this being I have been working with dowels the last week.

I’m making the interior frames to what is referred to as “the case” for a pipe organ. It houses the internals, wind boxes pipes so forth and so on.

A drawing of assembly sans the actual functioning parts of the organ.

7B07F7B6-12C0-4C6A-95E7-0B28141FEA61.jpg

I have only used dowels a couple times before to make passage doors. Honestly I was not impressed at the time as it felt slow and not precise at all even trying as hard as you could. Getting the two pieces to mate perfectly flush in any direction seemed impossible. Now close enough is easily obtainable but I’m not much into close enough.

So the new boss specified dowels to hold the internal frame together. The frame then gets wrapped in 7/8 thick bead board on battens and screwed to the frame.

60DE23BF-7696-4A79-9E2D-D9213015EB9F.jpg

912446BA-FFAA-4255-B10D-7675333F2AC7.jpg

And the front and back frame done. The four sides will be attached with countersunk #12 scored indexed on dowels or dominoes.

13642C5E-7EF8-4EE6-9F4D-B74AFFB6947E.jpg

This time around I found getting alignment side to side and front to back a bit more easy but still not even close to perfect. A quick sand does take care of flushing up any joints but honestly that’s not my style. I find having to do such kinda hackish and largely a big waste of time vrs getting the joint setup well off the machine. In the case of dowels I feel that’s impossible. At least with a hand held dowel jig.