I'm attaching a solid ash tabletop to the legs I've made. I have to install the table on site and don't want to drill holes to attach the ends of the legs. I was going to use typical wood buttons inserted into slots to secure the back edges of the legs.

My idea was to cut small wood wedges and slot the underside of the ends of the legs with a dovetail router bit so it works similar to a french cleat. When I pull the leg back it would force it against the tabletop. Would this method work or is there a better/easier option that allows for wood movement? Tabletop is 42" x 78" oval.enter image description here table render 1

enter image description here

enter image description here

  • 1
    I think you could do with diagramming the proposed attachment better as I'm having difficulty in visualising it properly so I presume I'm not the only one. I'm not sure I get your reluctance to drill holes when you're OK with routing the for slots? Another key question: the top is a 42" x 78" oval but which direction does the grain run? Only knowing this can expected movement be calculated. But we also need to know if the boards used are quarter-sawn or plain-sawn.
    – Graphus
    Commented Jul 19, 2016 at 8:26
  • The grain is running the length of the 78" direction. There is a drawing showing what I had in mind in the photos. Drilling holes would go all the way through the end of the legs which are much more visible than the slots which are hidden under the backside of the legs. Plus I would have to then fill the hole with a plug or leave the screw hole exposed, which is not ideal. This is all Flat-sawn boards. That would be an expensive tabletop using quarter-sawn! Commented Jul 19, 2016 at 14:28
  • Took me a while but I finally see how the various things function together. The semi-dovetail block I think will work as you envisage but you may want to increase the length of the slot and waxing the bearing surfaces might be a good idea, given the amount of movement possibly experienced between very dry and very moist. In a stable environment obviously little movement will occur, but best to plan for worst-case (which may be over 1/2").
    – Graphus
    Commented Jul 20, 2016 at 8:24
  • Offtopic: Very nice design! Commented Mar 28, 2018 at 12:56

2 Answers 2


To me, this design looks good structurally. (The table overall design looks great too!) As I look at your design it appears that you have thought this out very well. So I thought it might be helpful to our readers to discuss the key features your design addresses.

The dovetail connection will primarily serve to keep the table top flat and connected to the legs. It will experience the most stress when the table is being lifted or shifted. Lifting the table will apply force to pull the top and legs apart and the connection must resist that. The force will be applied to the beveled surface of the wedge and leg contact surface. IF the wedge is snug against the bevel cutout in the leg and is deep enough to prevent tear out of the wedge point under load, it will work. While you made the wedge hidden, I would be tempted to make the wedge wider than the width of the leg and both glue and screw it to the underside of the top to insure maximum resistance to the tension force applied during lifting. If the table is shifted or turned in place this will apply some lateral force on the legs and a twisting force between the top and leg assemblies (if the legs are still in contact with the floor). I believe that the combination of the buttons and wedges should be adequate to resist this applied force.

When the table is in place the load on the top will apply downward force across the leg assembly horizontal support arms. Since the leg vertical parts are angled the bottom of the legs will want to slide further out from the center. This will place load on the connection between the horizontal and vertical leg parts. The legs will attempt to close like a pair of scissors. The fact that you have made the connection deeper there acknowledges this. I would use full depth mortise/tenon connections to maximize glue connection between the two parts to resist the force.

Finally, if the legs resist the scissoring force adequately, which they should, the leg as a whole will try to rotate down at the center. The button connections (which allow for seasonal expansion/contraction of the wood top) near the center of the top will carry the brunt of this force as a tension force pulling the button away from the top. This force will be at least partially cancelled by the downward force of the table top, but some pullout force on the buttons may also be experienced when the table is in use. Make certain that the buttons are also strong enough to resist any pullout load both in the contact between the button and the top and in the slotted connection between button and leg.

One thing that is not clear in your images is that although you have a spacer between the vertical parts of the leg assembly, another spacer on the horizontal legs would be helpful to keep the legs rigid without depending upon the top.

  • Great commentary. I appreciate it. I actually have 2 cross braces on the horizontal member. 1 near the middle and one right at the end (close to the center of the table near the vertical/horizontal leg connection. I actually have the legs already made but all photos exceeded the MB size or I would upload them! I did use a full depth mortise and tenon for them and they seem strong. I really would like to avoid using a full width wedge if I can because I don't want to break the edge of the leg and have the attachment method visible. If the wedge is .5" wide do you think that would be sufficient? Commented Jul 18, 2016 at 14:46
  • I have another idea of using a keyhole slot with the router bit. I added it to the original photos. Much easier to do than making the slot I had asked about before. Attaching a screw to the tabletop and inserting those into 2 slots on the top of the legs (1 on each end, same location as the wood wedge was supposed to go). This won't allow the screw to move as the table expands, however. I don't know if the legs would be strong enough to prevent expansion and cause a crack. I feel like the ends of the legs would just expand the 1/8" or so that I'm expecting. What do you think @Ashlar ? Commented Jul 18, 2016 at 19:30
  • @ChrisHeichel I think the keyhole cx would work as well or better than the dovetail. I sympathize with not wanting an exposed connection at the outer end of the horizontal leg piece. I am a little concerned that the dovetail (or keyhole) connection could be too small. It occurs to me that if you provide a cross brace towards the dovetail location you might be able to expand the dovetail into a longer concealed french cleat if more connection length is required to support the loads.
    – Ashlar
    Commented Jul 19, 2016 at 1:20

That is a pretty sophisticated plan. The main issue you will run into is looseness. I don't really see how a simple dovetail is going to lead to a tight joint, especially if there is just one screw holding it. Tables have a lot of forces applied to them. People sit on them; lay heavy objects on them; stand on them to change light bulbs, etc. You don't want a flimsy table.

For the general design used here I would recommend sistering a whole board to the bottom of the table and cutting a tapered slot in the board. The entire arm of the leg would slide into the slot. The slot should be tapered so that the narrow part is innermost. To assemble the table the arm would be tapped into the slot using a mallet. This will make a tight, solid joint.

To attach the sistered boards with no screw, you can use hidden dowels or biscuits with glue, or just glue alone.

  • +1 for a dovetail batten, very similar to one traditional method for keeping drawing boards flat. But I don't think one of these can be attached safely using biscuits or glue alone.
    – Graphus
    Commented Jul 24, 2016 at 7:54
  • @Graphus You would be surprised how strong modern glues are. They are actually stronger than wood if applied correctly. Commented Jul 24, 2016 at 11:34
  • I know good joints are stronger than the wood around them, that's not the issue. Unless I'm misunderstanding what you mean by a sistered board that's attached cross-grain, which if glued doesn't allow for movement in the top. If the glue joint were strong enough not to fail the stresses in the tabletop that result will be relieved in another way, typically by cracking or bowing of some kind.
    – Graphus
    Commented Jul 25, 2016 at 7:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.