Rob, I can't help it. The Engineer in me wants to know how long an end grain joint will last compared to a dovetail joint. Remember a glue joint is a chemical reaction that can be affected by time and temperature. I would suggest sending both of your example joints to Texas and let them stay in an attic for a year. My attic can hit 120F easily on most summer days. I know you have a friend in San Antonio who would probably be willing to help. If not, I would gladly do the experiment here in Austin.
Keep making stuff with old world craftsmanship joints, but for items used for shop space you always have the shortcuts. Its kinda no different then people swearing by pocket holes and screws. Most tradesmen use pocket hole joints because they are fast and stand up fairly well. But you won't see a pocket hole jig in my shop. Its a shortcut. I have a 16x7 garage door all wood but its finger jointed short pieces. The glue joints over the 22 years have failed on the finger joints I think had it all been 1/2 laps they would not have failed or the bottom Rail is 1 1/2 × 16' x4" if it had been made out of a stack or wood plywood and wood with stager joints it would still be a solid beam. But no over its 22 years it smiles from tension of the lift springs. The only reason it doesn't pull apart is the metal brace. One of these days i remake each panel out of better wood.
I love how you are able to accept changes even though you have spent all you woodworking life "knowing" that end grain glueing is no good to finding out that it actually is good and immediately incorporating it into your work Excellent video thanks for sharing 👍
Excellent and useful video, Rob! I really appreciate how, as experienced as you are, you’re still able to be “blown away” by new evidence. You are a great craftsman with great character.
I love that you took Patrick's experiments and put them to the practical test. I also love that all of your channels are referencing each other. That is a strength of community not found in mainstream media. Well done again!!
I have to agree with Rob that cleaning up squeeze out needs to be considered in all cases. There's few things worse than squeeze out glue that defies cleaning up, in how it affects the appearance of the piece after being finished. Perhaps you're better at glue clean up than I, though, but I just haven't had much luck, regardless of the various methodologies I've tried over the many years I've been doing this.
That’s a great sign of someone who is open minded. Despite decades of doing it a certain way and listening to dogma, the guy is able to analyze new data, test the validity himself and make changes to his practice. Wicked job as usual Rob.
As a segmented woodturner I have always had to glue end grain to end grain to make my rings. Woodturners always thought the strength was in gluing the long grain of the rings together and using a brick lay pattern. Patrick's video demonstrated to me that my segmented bowls were very strong if the glue was applied correctly and the rings are flat. I shared his video with my woodturning friends because it just verified what we're doing on a segmented turning. With flat work the issues are different and you've done a great job of demonstrating the impact to your joinery. Thanks Rob!
I first encountered this phenomenon in 1970, when white glue was still commonly used in woodworking, even by professionals. Titebond glue was a revelation to me, as the breaking strength of the glue was so much higher than that of white glue. I bet that performing these tests with older glues will that end grain joints are indeed inherently weak, depending on the kind of glue used.
An interesting point, and that would make sense to be true. White glue is significantly different, and it may even be that the end grain wicked out the moisture from the glue so fast that it didn't properly polymerize as it set?
Just be careful of your first test with the dovetail and rabbet joint, if you hit the rabbit joint with a mallet (like the dado test) it does in fact fail. The dovetail does not. I was after something that was consistent so the test I performed was as follows: 1) Clamp the job to a post so that the part to be put under load is poking out. 2) Using two house bricks taped together drop them onto the section poking out as the same distance out from the job and at the same height (I used 8 ft). 3) Record results. My result may be slightly different to yours but you could very well find that the weakest joint is not the dado.
Funnily enough I started woodworking when I retired and made some cabinet drawers with glued rebated corners not knowing that this was not a thing to do. My drawers are still hanging together solidly so my estimated necessary rigidity for my drawers is now borne out scientifically. Thanks Rob, Stumpy and Patrick.
It's really just a testament to how far glue technology has come over the years.. easy to forget how much chemistry and science has gone into these products.
No, hide glue is stronger than any of the glues we regularly use in woodworking, some of it is used to chip glass to make fancy designs, try that with titebond.
But dont be fooled by a scientific experiment tailored to get a result that bears no meaning on practical woodworking! . It doesnt matter how superior the wood glue is today, the lignum that bonds wood fibres together is the key factor in strength (demonstrated by Mr Sullivan) and centuries of wood jointing respect that overlapping/extending fibres give strength. Wood is not like metal.
Someone showed me years ago a wonderful trick for cleaning wet glue out of those tight corners. The opening of a regular plastic straw can be easily pressed tightly into the corner edge and scoops the glue out cleanly. When the opening fills up, snip the end off and repeat. A single straw will remove a lot of squeeze out glue.
More than 50 years ago I was taught basic woodworking by my uncle. Every joint, no matter what the orientation was, GLUE IT. My uncle learned from old masters. It was hide glue or some form of glue, but, he said you can clean area after so DO IT. Thanks for the info Professor Cos, my uncle would be proud.
Rob, I always say that the best thing that can happen to me is to find out I have been wrong about something, because two wonderful things have happened. I have learned something new and I have been given the opportunity to stop being wrong. I love both. I watched all three of the other videos before I watched yours-I actually sent you an email about the issue. Other than the original, your was the most useful.
Thanks Rob, I watched Patrick's video and his reasoning made sense, the tests he did fortified the reasoning. You've added to it with real project application. Pretty cool.
As a Materials Engineer, I remain skeptical. In my opinion these one time static tests are only part of the equation. We need to assess this in a more comprehensive scenario by including cyclical racking and fatigue to see if the durability of these joints across the system of joints used in a piece is sufficient long term. Mechanical locking combined with glue surfaces of all types has the best chance of surviving long term use, wood movement, and the ubiquitous moving of furniture from room to room and house to truck to house over the course of generations. In the design of a piece use this knowledge appropriately by having critical joints absorb the racking forces with mechanical aid and other joints that typically do not see raking to use only glue surfaces (i.e. outside corners of a book shelf using dovetails and all the non-moving shelves using dados only). I am also one of those people who spends way too much time scraping glue out of my damn corners...
Of course you will also have to test several differnt types of glues. Looks like Rob is using something other than Titebond III which is very popular with many other woodworkers.
This whole knowledge of connecting the long grain end to the long grain end is actually incorporated into some Japanese joinery like a sleeve I think is the name of what I'm thinking of it's a long cut construction joint to connect two beams together then forced into each other with a wedge but it's connecting end to end and it's one of their strongest joints used in traditional construction edit:after thinking about it then finally just googling they use a ton of end to end joinery in traditional Japanese wood work they even have wats of turning a joint into ends touching ends when it doesn't have to be they go out if the way to join end grain I had never thought about it before. I know their whole culture and mindset in making anything is to use physics and nature to your advantage and work with it instead of fighting against those forces as we see it in the western world and those two different schools of thought are so obvious in everything we do honestly. I can help but look at some of their incredibly complex end to end joints and think God that's smart but it mimicks the tree itself by joint a long grain interlocking with another long grain then forcing them into each other as gravity would and giving them some flexibility you've mimicked the tree and maintained it's strength. We cut stuff apart then try to force it to have some strength again by glueing it to a strong piece we haven't cut through yet. We fight nature they use it.
Millions of craftsmen working the same wood for thousands of years with nearly the same tools and we are still learning new things. One of the true joys of wood working.
there will always be a need for proper joinery, but with these revelations that have come forward from not only yourself, but some of the other channels that I subscribe to as well....this, for some applications, is going to be a game changer. I appreciate the effort put forth with a few various types of joint you did and "tested", its easy to see that in certain spots with end grain versus long/face grain, a little glue will absolutely help with strength even if it is on already classically strong joints to begin with.
Never would i ever built a corner and not put some nails in it. I glued everything before, now no nails or screws. I have to build a bookshelf with only glue, even the back panel, i need to know. Thank you for this eye opening day Rob and Mr. Sullivan! Holly ....!
Everyone's been talking about glue surface and wood properties, glue properties the science of wood joints. Throwing around terms like strong, stronger and strongest. Rob is the only woodworker to post a video checking on strong enough.
You are great Rob. You did this video, challenging yourself and investing your valuable time and materials. The Best of this, is that you did not do the video trying to "explain" Mr Sullivan's work. You earned my subscription Rob!
Awesome video Rob! I too couldn't believe how strong that rabbeted joint held up. As you mentioned, the dovetail joint is more aesthetically pleasing, but for a rabbet glue up to be that strong is mind blowing! Makes me ponder things moving forward!
Rob, for me the critical problem is “TIME”. I’m Italian, in my home I have some 16th and 17th century pieces of furniture. Dovetails, dovetails and dovetails. Sometimes some rabbet or dados with forged nails. Do you think that in the centuries the rabbet joint that you tested can resists? Large pieces like these have a big tendency tu cup and no way for the glue joint. Dovetails last forever.
The benefit of dovetails is that they function even without glue. The rabbet lasts only until the glue fails, so it really just depends on how long we think the glue will hold before it becomes brittle and breaks.
@@nickglover exactly. Sincerely I have no idea how long glue can performs. I suppose it depends on the environment, moisture, exposure particular situation...a chest of drawers near a fireplace can be stressed a lot in the wintertime. UV is another important fact in some regions.
Another great response to Patrick's great video. If anything it just shows how lucky we are that we have the glues that we have that are stronger than the wood and it is just sitting on your shelf ready to be used whenever.
In addition to your expertise, I've always enjoyed your clear and precise way of explaining your concepts. This was a great transition from scientific to practical.
It always amazes me the way Rob attacks woodworking with the skill of a heart surgeon. I know tools can play a part, but his skill is the main tool in his arsenal.
Hey Rob, regarding glue cleanup from difficult places: a little bit of diluted shellack (or any other finish that is planned) on potential squeeze out areas will make cleanup a breeze. A plastic pick or something will make glue just pop off. The downside is a very careful application is required so that glue surfaces are not affected, but after a bit of practice padding some shellack is quick. Shellack is also easily removed, almost completely.
Thanks Rob. A great video which wonderfully complements the videos from Stumpy Nubs and the Wood Whisperer. Between the three of you, you have covered the topic comprehensively.
Excellent follow on from Patrick's video. Most traditional woodworking was developed prior to the development of the excellent adhesives we now use. I can't think of a PVA glue joint I have made that has failed in 'normal' use.
Your practical application & testing of Patrick's results are the perfect pairing and demonstrates the possibilities. I really like how you describe exactly how and why you do things.
Being a mechanical engineer I have to rise some doubts about the forces applied to the pieces with leverage and hits and loads. Sullivan used a more standard testing method that is the core of his research. I don't know if the two results could be comparable. After all, this video reason was different. I'll ease up things by doing the easiest joint next time and see how much it lasts. Anyway glues has improved a lot. Thanks to all the contributions on this task!
I think it is very important to remember how pathetically rubbish wood glue used to be! If I had made that exact joint in my woodwork classes at school, it would have broken just by picking it up and shaking it. LOL So the wisdom of that era was correct and undisputed. But now with glue’s being stronger than the wood itself, it stands to reason that it can make a better joint than back then. How long that joint remains strong and reliable is to be seen, especially if you take into account environmental conditions and cycles of work or loading, and the position of the joint. I’d like to thank Mr Sullivan, Stumpy Nubs, The Wisperer and yourself Rob for all the videos and the integrity and character of all of you. For now at least I will still be reinforcing my joints or using dovetails.
With the dado / housing: When cutting these by hand, you cut the sides, clean the middle, and refine with a router. If when cutting the sides, you go past the depth by even the smallest amount, then the bottom of the housing will be short pieces of face grain only as long as the width of the housing because of the fibers being scored along the length of the sides of the housing. Maybe this might be what is limiting the strength in your test? It looked like you had a clean line of torn out face grain fibers adhered to the end grain of the shelf...no ragged edges. That's what made me wonder if the bottom of the housing was really short (face) grain. The torque on the side walls was clearly enormous because it just pried off those pieces below the shelf. Thank you for sharing all these tests. Very interesting. I was sure that rabbeted corner would fail. I'll bet that if you do a pure tensile test on the housing, it is strong enough for tensile loads. The back of the cabinet could take care of the racking loads, perhaps?
I took it for granted that people had already tried Rob's experiments over the years and found them lacking. Its more surprising to me that these ideas have just been accepted as fact by the general community without verification. I can only assume that this is partly attributable to the quality of modern glue and the amount of time that its been available to woodworkers...
Like most, I've always believed end grain glue joints to be weak. I think what Patrick's video (and the subsequent follow up videos such as yours), has emphasized is that a joint is only strong as it's weakest link. What's become apparent is that most of the time, It's not the end grain glue joint itself that's weak, it's the lignin bond that hold the fibers together. In cases where an end grain glue joint does fail, it's only because cross grain is incredibly strong.
The stresses and strains expected in furniture use are taken into consideration at the design stage. It is called joinery for good reason. All the main stresses are covered even if no glue was used at all. In most construction all the glue is needed for , is to provide enough strength to stop the joints coming apart. All the main tresses are carried by the wood alone. We have known for a long time that glue is usually stronger than the wood, what ever the direction of the grain. However there is no actual gain in this if the wood fails first. More important is choosing the right glue for the expected conditions. A beautiful veneered desk will soon delaminate if left out in the rain, or in a damp cellar. Not all glues are suitable for all purposes, and in the toughest conditions such as timber framing no glue is usually used at all. End grain to end grain is not something any one needs to use or even worry about very often and in most cases is best avoided. However commercially available timber and boards are now often spliced end to end. And are exceptionally strong.
Years ago I watch a video of a woodworker who explained that the end grain glue up was weak because the end grain fibers would suck up the glue and starve the joint. At least that was his take on the problem. His solution was to prime the end grain with a little glue first and allow that to be sucked up and then apply the regular amount of glue to the joint.
Nice to see you make this video Rob....three of my favourite youtube woodworkers all giving comment and now a practical application test too.....best thing to see here?.....you still being able to be surprised by a result you in all honesty did not expect....a day with learning something is always a good day :)..weldone mate and thanks for taking the time to do this :)
I've always believed the myth also. Pretty interesting. As to the lap joint not failing..... pretty good. If the joint had been a butt joint, it would have failed, but the fact that the dado was fairly deep, then the the stress was relieved quite a bit by the "shear" strength of the base of the dado. On the other hand, the dado joint, (for the shelf) was fairly shallow. IF you had made that one deep, it might well have withstood your test. Still though, the possible stress on a shelf is normally pretty directly downward, and thus, even a shallow dado is probably fine, AS LONG AS there is a back on the shelf to prevent racking.. thanks Rob
One thing to consider with the rabbet joint though is that the glue is in the place where the wood will move seasonally. So if the glue doesn't flex it may separate from the wood itself, and that may cause the joint to fail earlier than the dovetail joint, which will stay strong longer since the way the glue is positioned moves way less.
The dado joint (last piece in your video) is failing along fibers but not on fibers. The details are a matter of forces and its distribution inside the wood. That's part of stress calculation in direction of the fibers and perpendicular by an engineer. If you angle the glue surfaces to equalize stress along and perpendicular to the fibers proportional to the wood strength of both directions you're getting the maximum strength for a joint ... that's what makes finger joints so strong.
Awesome video Rob. I myself always make little test when I'm curious about the strength of a joint. Most of the time I'm pleasantly surprised at how little it takes to be be structurally strong "enough" for my application
I watched the response vid by another woodworker, he basically pointed out that the test was correct in a lab setting, but in terms of woodworking, leverage still means it is a weak joint.
Lets be honest: the dovetale joint, along many others, is not used because of its mechanical strenght, but because of its appealing aesthetics... Still, it's good to know that the bond in more simple joints appears to be sufficient.
This may revive dowel joinery, but I find dowels always failing on the end-grain to long grain contact part of the dowel., not the long grain to long grain part of the dowel. This is why I have shunned dowel joinery. This may also entertain the idea of gluing plywood edge to edge, since there will be end-grain to end grain contact in a part of the join.
I built a display case in the Barrister fashion for my T-rex jaw piece and I used a rabbit joint on the two sides of that oak case to fix the sides to the base. For years I've regretted not using a locking rabbit joint. Now, I know it's a cosmetic choice I'm regretting, not a structural choice I neglected.
That's what Luther said too!!!! I wont be throwing out dovetails for rabbets.....There is just something about the pleasure of cutting dovetails and the look of quality about them
I am with you if i didn't see it i wouldn't have believed those 2 joints could withstand that force of you standing on them. I am as stunned as you, tthey didn't even appear to flex at all. I know Dovetails are strong but the rabbet as well!?
Mr. Sullivan blew up woodworkers youtube! This has been informative and entertaining. I think Mr. Sullivan has explained why scarf joints are so strong
As I was watching, two questions came to mind. One, you addressed by shooting the end grain versus leaving it rough from a tablesaw. Two, I wonder if sizing, and /or, applying thin amount of glue to end grain, let it dry, then apply more glue before finally assembling and clamping? Hmmm...great video Mr. Cosman. Thank you.
Wood glue essentially is pulled into the fibers of the wood and solidifies to itself. So it's bonding to itself, not the wood. If you let a layer dry it can't "seep" into the wood so you'll have an extremely weak joint. In this case it seems it had a pretty deep penetration. So the results may vary depending on the type of wood itself. The more porous the wood, the better the bonding.
I like the application of his testing. Nice work! Patrick’s testing showed the end grain glue joint is a bit stronger than wood in its weakest direction. This gives a practical application of this. I have always glued end grain since it did add some strength and is easy to do. I was surprised with the results that it is stronger than I had assumed. As you indicate it doesn’t change most joinery methods but it’s good to see there is some change in applications and this bit of joining knowledge makes us all a little better.
Rob, thanks for sharing this important revelation with us. I think that it would be unrealistic to believe that the 1/4" deep dado joint could be as strong as the much more complete glueing surface of the rabbet joint. Very well done, Sir.
Hi Rob I have great respect for you and now your willingness to adapt. I discovered your channel sometime maybe 2 years ago and without a double you have changed my woodworking skills more than any other one person. This video has gone most of the way to close the gap on the few things that I have disagreed with you about. 1) End Grain gluing 2) That two pieces of wood need to perfectly touch in order for a joint to succeed. I am not advocating poor workmanship. I am advocating that at the bottom of a joint if it hollows out a bit that that joint will actually hold better than a perfect joint. I know this sounds crazy but from some testing I have done in the past a joint with a small glue hollow will hold better than a perfect joint. There is another channel that did some glue experimenting that also seemed to agree with my theory. The primary reason I bring this up is your dovetail hollow, not that that joint needs more strength. Not asking you to prove me right or wrong or that you should change your practices. Thanks again for all that you have done for my skill level.
Thank you very much, watching this video was a pleasure. It would be nice if you broke two similar-sized joints (rebate and dovetail) and compared their strength (simple lever and a water can works just fine for this).
You can glue two 2×2's together end by end, you just need to make a scarf joint, slope should be 1/12 to 1/15. It is used in aviation for structural joints.
Perhaps we should not forget that scarf joints do exist. Patrick's video was excellent, but many people have missed the point somewhat. In Patrick's data, what really mattered wasn't comparing the side grain to end grain glue joints, but rather the end grain glue joint to no joint at all.
Rob, I was taught in gluing end grain, to first prime the end grain with glue. As end grains make up is as porous as it is, this technique allows the wood to absorb the initial glue. Wait a couple minutes after priming, then glue as you would ordinarily. This no doubt will strengthen even further the end grains ability to adhere to its mate. I've found this to be true more than expected. What are your thoughts?
The only question I have is, 'does the strength of the end-grain joints hold up the same after a few years when the glue has aged? Typically, after 25 years or so, glue deteriorates and starts to fail so the end-grain glue-ups would fail drastically whereby the dovetail ends would still hold fast after 25 years even if the glue deteriorates or even without glue sometimes. All this is good to know , thank you for showing this test.
I think the main problems with the shelve is the leverage you used and the way the dado is cut. You're very likely to cut the dado deeper on the sides than the middle, leaving only a very short piece of long grain attached to the upright board. The most likely failure of a bookcase is when the shelves are so loaded they either sag so much they pull inside, or there's so many books stuffed on the shelves they wedge the sides apart. Just gluing end to long will not be strong enough for those scenarios, but dado's or shelve supports will prevent a large part of those failures. Besides, you usually have 2 sides supported, making the leverage issue non-existent.
Hey Rob. I actually make all my drawer boxes with that rabbit joint and they stay together forever. I have many draws that are 40 years old that haven’t moved at all.
my experience has been with glue drying out over a period of years along with wood movement. a year or 2 with but joints and up to 5 to 10 years for mortise joints. but with a mortise joint there is the mechanical connection.
I saw the Wood Whisperer's video, which was interesting and informative, and seeing your methodologies and explanations were just awesome as always! Interesting results, for sure.
Absolutely great video!!! Well thought out and extremely useful. The one joint in my mind is the 45 degree mitre end grain to end grain. It still needs reinforcement? Thoughts??? 💭
Rob, one idea. You should clamp(or fasten) the rabbet joint and the dovetail joint on the one side together and then apply force to the other arms until one of them breaks. I bet that will be the rabbet joint.
Yes, that would have been a far more accurate test. You could put a couple screws in that rabbit/rebate and no glue and on an angle like that it probably would give the same results. The forces on the joint surfaces at that angle, hold the boards up. We don't stand a cabinet or case on edge, but we do put weight at 90 degrees angle to the joint. Also, there are raking forces on the joint that standing on it doesn't really test.
This video is delightful. Everyone should be as open to new information and to empirical evidence. The thing I most appreciated, though, was the thoughtful analysis of the practical meaning - yes everyone has been wrong about end-grain glue strength, yes it has a few meaningful ramifications for actual work, but no it doesn't mean joinery is obsolete. When we learn, we adapt - we don't simply throw away what we've learned before. My takeaway from Mr. Sullivan's original video was "PVA glue is stronger than lignin but much weaker than wood grain. If maximum strength is needed (and due to leverage it often is), try to make a joint failure force a failure of the wood grain rather than accept a failure of the glue OR the lignin. That said, stop worrying so much about face grain vs. end grain for glue surfaces - that's not really what matters."
Great video Rob. I really appreciate the practical application of the information from Patrick’s video. With respect to the dado joint I have always used a sliding, tapered dovetail to provide the mechanical strength because I assumed the end grain to long grain connection would be too weak to prevent the sides from bowing out. I guess I will have to re-evaluate whether I still want to do the tapered, sliding dovetail. It is more work and more time. It would be interesting to figure out a way to measure the difference between a regular dado and a sliding dovetail. When my new shop is completed I would like to see if I can figure out a way to do that test and film it.
That rebate joint was a surprise. I was surprised as you Rob. I have restored antique and period furniture for 40 plus years and know that time is a big factor that is rarely factored into the equation of these tests. Can I suggest you retest the dovetail and rebate joints periodically over an extended period of time. Regards, Terry Gleeson. Sydney
Prime the end grain first with glue. Wait til it's been just absorbed, then glue as you normally would. This will guaranteed make even more of a difference, especially with the dado. david
One thing I don't understand. In the half blind dovetails, why didn't you always glue the end grain of the dovetails to the face board? Even if you thought it would be a weak bond, a weak bond is still better than no bond.
I think standing on the joint puts less pressure on glue and more on the woods fibers. I think putting weights only on one board will show the glue's strength vs the wood fibers.
Thanks for a sincere and through video. I'm impressed you are so thoughtful which faces get glue in a joint. A true master is in the details. It seems you'd have to intentionally not glue the end lap. Was there ever any reason not to glue it?
Says Rob (about the shelf case end grain glue up), "it didn't have the shear strength we look for but it was very strong". Rob, you didn't measure the "shear" strength of that construct. Shear is hard to measure without specialized lab equipment, but if you had measured it you would have found it is a lot greater than you imagine. As for Patricks findings, that has been known to be the case for a very long time and is documented in research documents freely available.
