The Anvil Test
It has been said that we learn more from failure than from success. One of the first pieces of furniture I made was a coffee table with rails tenoned into the legs. I didn’t know that the tenons were too large in proportion to the size of the legs until several years after the table was finished.
On the way to the bathroom one night, I accidentally kicked one leg. The wood surrounding the tenon gave way, and the leg went flying. Up to that point, my work was successful; yet in an instant, a nice table was reduced to kindling. This was a joint-design lesson learned the hard way.
Much of what we know about woodworking is someone else’s experience, or the embellished retelling of someone else’s experience. When we find a method that works, we hang on to it and then we recommend it to others – never sure if it really is a good way or if we just got lucky.
If you want to start an argument, or at least a lively discussion among woodworkers, the relative strength of joints will almost always get one going. Like most good arguments, this one can’t be settled conclusively.
In order to add a little fuel to the fire, we decided to test 10 common joints by dropping an anvil on them. We used two sizes of anvils: 26 pounds and 54 lbs. All our joints were carefully prepared using poplar, and they were glued together with Titebond Xtend glue, which was allowed at least 24 hours to cure. The anvils were dropped from about 18″ above the joint using the contraption shown above.
We picked an extreme example not to crown a king of woodworking joints, but to examine how and why joints fail. After deliberately destroying these joints, we found ways to improve our joinery methods and produce better work.
Weaker Carcase Joints
The first joints we examined were simple butt joints, and some typical ways of reinforcing them. All of these broke beneath the 26 lb. anvil, a result we expected. What we didn’t expect was how well the polyvinyl acetate glue held.
Most of us have been taught that glue will not hold on end grain, and in the mitered joint, the glue did indeed fail. The grain in a miter joint is mainly short-grain to short-grain.
In the rabbet joint, however, the bond is long-grain to end-grain. Does the presence of long grain improve the strength of the joint? As you can see, the wood fibers failed beside the glue line. If you look closely at the photo, the end grain of both pieces isn’t seen. Long grain from the adjacent piece is.
This was true with all of the joints we tested. From a simple butt joint to through-dovetail joints, the glue typically held to the long-grain wood fibers, even when the other component of the joint was end grain.
In the housed rabbet joint the wood broke neatly across the joint line, leaving a thin layer of long grain intact. The break continued across the short tongue that held firmly to the end grain of the two adjacent surfaces. This joint was theoretically stronger mechanically, but it reveals our second important discovery: The wood will break at the weakest point within or near the joint.
This is important when designing joints. The sizes of the components of the joint are important, but so is the amount of wood next to the joint. The mortises and tenons on my coffee table weren’t bad, but there wasn’t enough wood left beyond them for the joint to stay together under stress.
The biscuit joint above held together where the biscuit was in contact with the wood, but the wood itself broke out at the short grain at the edge of the biscuit. The weakest point wasn’t at the joint; it was just beyond it.
Stronger Carcase Joints
As we expected, box joints and dovetail joints were significantly stronger than the first group of joints. They each survived the impact from the smaller anvil with minimal damage. With the larger anvil, the dovetail at top gave way but the two pieces did not completely separate. Notice that the glue held on to the end-grain surfaces, and that the wood fractured on the short grain of the tails. The joint opened up about halfway, where the tails and pins wedged against each other.
The parallel surfaces of the box joint above, without the benefit of the dovetail’s wedging action, unfolded like a hinge. Once again, failure occurred from the wood fibers at the glue line, not the glue itself.