george nakashima spindle detail

[Art Brazee]

I'm building a bench after George Nakashima's benches and chairs. Mine is somewhat different, but I am using his spindle/back support detail. His backs are supported by a triangle; regular spindles in compression supported by two outside spindles in tension. The two outside spindles are larger and attached forward of the regular row of spindles producing a greater angle of attachment.

The mystery is this - pictures of his work (and I've looked at many) show a conundrum. It appears that both the outside spindle and the regular attach parallel to the angle of the back. 271533This is impossible. Further, the outside spindle goes through the back. I don't believe the rake changes because of the curve. Many of his backs are almost flat or with a very slight curve. I think I many end up shaping the back after it is assembled, but maybe someone's tackled this detail before.

I've spent a long time looking at pictures so guesses won't do. I'm looking for someone who's worked such a detail in a similar piece. 271534

[Prashun Patel]

This is not a guess, but two questions:

Does the bottom edge of the back rail remain in flat in the same plane along it's length?
are the tops of each 'regular' spindle coplanar or is there a slight arc?

[Art Brazee]

This is not a guess, but two questions:

Does the bottom edge of the back rail remain in flat in the same plane along it's length?
are the tops of each 'regular' spindle coplanar or is there a slight arc?

Great questions...

1. Generally yes...it appears that some are flatter than others, but I think I can say that the plane, if broken, doesn't equal the difference of the rake in the spindles. If the back is raked the rack is very slight.

2. Looking at his work, I can't pick up an arc. He did a great job hiding the solution - thus, his genius. If there is an arc he had to accomplish the change is about 3" of length on the back.

I think you immediately drew in on the big question. If I imagine the back, bottom edge scooping up and forming a differing, ever greater angled plane each spindle would point more toward the center of the arch and higher. If I were to angle the outside, supporting spindle an additional 10 degrees, that 10 degree change projected on the back would be substantial. I just don't see that in his work.

[Pat Barry]

I don't understand your comments about tension and compression, but, it seems obvious from the second photo that there is a curvature to the end of the back where the larger spindle is attached and it sure seems to be curved both forward and upward (slight twist). Are you sure this isn't what is going on? By the way, where did you get the pictures? Did you take them? If so, a couple quick measurements would answer all the questions.

[Art Brazee]

I don't understand your comments about tension and compression, but, it seems obvious from the second photo that there is a curvature to the end of the back where the larger spindle is attached and it sure seems to be curved both forward and upward (slight twist). Are you sure this isn't what is going on? By the way, where did you get the pictures? Did you take them? If so, a couple quick measurements would answer all the questions.


I've not seen a Nakashima piece in person, only pictures. Few of them really show what I'm interested in.

Regarding tension and compression. Nakashima's benches are pretty minimalistic and the back doesn't have much support to prevent it from flexing and perhaps breaking. The only real support is the end spindles which are in tension. When the back is weighted, the end spindles are pulled. He created a triangle, if you will, which supports the back. Take a look at some pictures of his benches.

[Jerry Miner]

Ok--you piqued my curiosity. I've never tried to duplicate a Nakashima bench, but I fooled around with SketchUp a little and I created the attached drawings. (SketchUp is great for exploring the geometry of furniture).

271630271631271632

There are a couple things at work here (at least in my drawing. It would be interesting to take a close look at an actual Nakashima bench):

The top rail is tilted back a little more than the standard spindles.

The end spindle is tilted slightly more than the top rail (the undercut of the top rail helps to hide this.)

Canting the end spindle sideways a bit will also help hide it.

Bottom line: it is not impossible, but it's also not as straight-forward as it looks.

[Art Brazee]

I wish I could pull off drawings so easily. I did a drawing, but the complexity bedeviled me. You're right, it's not a easy as it looks.

Here's what happened for the benefit of others. In order for the side spindles to pass through the back the back really has to be at the same angle as the spindles. I didn't find a way around that. However, I made my regular spindles at the same angle as the back. So, I didn't pass the support (side) spindles through the back. My back is larger too, so it was problematic anyway. Here's the result.271764271765271766

I obviously made my bench somewhat different, but it certainly is a child of Nakashima. I think his genius is that, at least from pictures, you can't tell what he did. However, I did find this picture after finishing. It tells the whole story.271767 I would like to sit in one of his benches so I could feel the back. I would think the missmatched spindle and back angle would be odd, but I doubt that's the case.


Ok--you piqued my curiosity. I've never tried to duplicate a Nakashima bench, but I fooled around with SketchUp a little and I created the attached drawings. (SketchUp is great for exploring the geometry of furniture).

271630271631271632

There are a couple things at work here (at least in my drawing. It would be interesting to take a close look at an actual Nakashima bench):

The top rail is tilted back a little more than the standard spindles.

The end spindle is tilted slightly more than the top rail (the undercut of the top rail helps to hide this.)

Canting the end spindle sideways a bit will also help hide it.

Bottom line: it is not impossible, but it's also not as straight-forward as it looks.

[Pat Barry]

WOW! Really nice job Art. I didn't realize you were so close to finishing. It turned out great.

[Alan Caro]

Art Brazee,

It's certainly an interesting design. I don't have direct experience with the Nakashima design, but my tendency would be to think of it in the terms of Windsor Chair backs- a rigid compression arch-though the Nakashima is extremely shallow- a flattened Windsor supported on levers. In traditional Windsors- similarly a quite strong design made from ostensibly lightweight components-the force is distributed by the rigid back fairly evenly over all the spindles and the spindles curved mounting pattern provide a moment resistance by an increased offset angle- they are restraining elements. It's perhaps confusing to think of the general spindles as being in compression. In my view, they are are mostly acting as levers, restrained by the end spindles angular resistance- in this I'd refer to the general spindles as flagpoles. The only components I would consider are in tension are the end spindles as they are restraining by the angular displacement. Yuo could imagine that these could even be cables - pure tension elements if the angle were increased enough. And, as in flagpoles, another aspect of strengthening is to consider the thickness of the bench bottom / seat, as the greater the proportion of embedment of the general spindles, like a flagpole being deeper in the ground, the greater the moment resistance. This resistance could be substantially increased if the field (general) spindles were arranged in even curve and more spindles had an angular displacement at the base- thereby making a more even distribution of the horizontal force load. Again the Windsor Chair is the model- the more like a rigid arch arch distributed over the greatest number deeply embedded flagpoles and the greater the curve of the mounting- the stronger it is. An elegant design as the obvious solution would be to have the end spindles angle as far behind the field spindles as a brace- imagine them at a 45 degree angle or even alternate of near vertical and 45 backward bracing- and that would be an effective use of a compression element. There are backward-angle (compression) bracing rarely in this structure, but sometimes seen in the more cautious form of Windsor Chairs - made for "substantial" clients perhaps,...

Here is a modern interpretation of a Windsor Chair in bench form by Chris Harter and I think it clarifies the Nakashima concept, by simply having more of the multi-angular restraining elements distributing the load along a rigid arch.

http://www.chrisharter.com/windsor_settee.jpg

Alan Caro

[Art Brazee]

Alan,

Thank-you for the terrific and elegant description of what's actually going on. I was concerned about the strength of the back; however, upon completing the bench I realize I could have made the spindles with a smaller diameter. You explanation tells us why. Mine is radiused also, but the picture you included certainly does show the restraining elements. Having made my first bench, I have even greater appreciation for Harter's work - wow, tremendous complication...so many angles (restraining elements).

The picture of the bench you included
Art Brazee,

It's certainly an interesting design. I don't have direct experience with the Nakashima design, but my tendency would be to think of it in the terms of Windsor Chair backs- a rigid compression arch-though the Nakashima is extremely shallow- a flattened Windsor supported on levers. In traditional Windsors- similarly a quite strong design made from ostensibly lightweight components-the force is distributed by the rigid back fairly evenly over all the spindles and the spindles curved mounting pattern provide a moment resistance by an increased offset angle- they are restraining elements. It's perhaps confusing to think of the general spindles as being in compression. In my view, they are are mostly acting as levers, restrained by the end spindles angular resistance- in this I'd refer to the general spindles as flagpoles. The only components I would consider are in tension are the end spindles as they are restraining by the angular displacement. Yuo could imagine that these could even be cables - pure tension elements if the angle were increased enough. And, as in flagpoles, another aspect of strengthening is to consider the thickness of the bench bottom / seat, as the greater the proportion of embedment of the general spindles, like a flagpole being deeper in the ground, the greater the moment resistance. This resistance could be substantially increased if the field (general) spindles were arranged in even curve and more spindles had an angular displacement at the base- thereby making a more even distribution of the horizontal force load. Again the Windsor Chair is the model- the more like a rigid arch arch distributed over the greatest number deeply embedded flagpoles and the greater the curve of the mounting- the stronger it is. An elegant design as the obvious solution would be to have the end spindles angle as far behind the field spindles as a brace- imagine them at a 45 degree angle or even alternate of near vertical and 45 backward bracing- and that would be an effective use of a compression element. There are backward-angle (compression) bracing rarely in this structure, but sometimes seen in the more cautious form of Windsor Chairs - made for "substantial" clients perhaps,...

Here is a modern interpretation of a Windsor Chair in bench form by Chris Harter and I think it clarifies the Nakashima concept, by simply having more of the multi-angular restraining elements distributing the load along a rigid arch.

http://www.chrisharter.com/windsor_settee.jpg

Alan Caro

[Yonak Hawkins]

I look at this Windsor bench and think of the Mel Gibson character in The Patriot. I wonder how many times the original designer of this chair sat in it, leaned back and it collapsed before he finally got the design technique right.

[John Pond]

Great looking bench. I am attempting one myself and quickly realizing it will be more challenging than I expected and possibly beyond my skill set. I have several questions for you if you don't mind; 1. What thickness did you use for the back rest? 2. How deep did you enter the spindle into the backrest and also the slab? What kind of drill bit/method did you use for drilling to avoid tear oout at the edges? What angle did you use for the spindles and side support, and why did you choose to match the backrest angle to majority spindle angles? Is the back rest curved along the main spindle portion, and if so is the spindle line where it enters the slab a straight line with slightly differing angles? And did you steam the back rest to achieve the curve, and does the curve increase towards the end? Any steaming referencing you could recommend would be helpful. If you could help with any of those it would be greatly appreciated and I will include some pictures of this beauty slab I am working with to sweeten the deal. It is a spalted maple slab that I milled a couple years ago and for some reason dried quite nicely, not much movement in the wood, maybe to do with the spalt. Thanks Alot317574317573317575
3. Did

I wish I could pull off drawings so easily. I did a drawing, but the complexity bedeviled me. You're right, it's not a easy as it looks.

Here's what happened for the benefit of others. In order for the side spindles to pass through the back the back really has to be at the same angle as the spindles. I didn't find a way around that. However, I made my regular spindles at the same angle as the back. So, I didn't pass the support (side) spindles through the back. My back is larger too, so it was problematic anyway. Here's the result.271764271765271766

I obviously made my bench somewhat different, but it certainly is a child of Nakashima. I think his genius is that, at least from pictures, you can't tell what he did. However, I did find this picture after finishing. It tells the whole story.271767 I would like to sit in one of his benches so I could feel the back. I would think the missmatched spindle and back angle would be odd, but I doubt that's the case.

[Art Brazee]

I believe the back ended up about 3/4". It looks too heavy at a greater thickness.
For the most part, both were mortised about an inch, as I recall. I used a forstner bit on the drill press with blue tape which held out the layout lines. The layout sweeps with the curve of the back which is gentle with a sharper return on the ends.
I had not made a chair back before so I spent a lot of time looking at what angle others used. I've forgotten the angle, so I'll have to measure it. I think it was 12 degrees. I used the same drill angle for the spindles and the angle of the back; i.e., the drill angle went in at 90 degrees to the back rest. The back rest is curved in both directions. I cheated finding a very curved, but nicely sawn "reject" in the walnut pile which was perfect. It has almost no grain runout and I shaped it by hand. I further sawed it and curved it. The sides sweep up. I don't have any access to stack dried walnut, so steaming was out - it was either lamination or finding a nice curved piece.


The only two spindles that are at a different angle are the two large spindles. These form a triangle which support the back. This works so well the back is unmovable - there is zero flex. I was quite surprised. The end spindles pass through the slab and are wedged through the bottom. The other spindles do not pass through.

Good luck...let me know if I can help. I learned a lot on this piece and was happy with the result.


Great looking bench. I am attempting one myself and quickly realizing it will be more challenging than I expected and possibly beyond my skill set. I have several questions for you if you don't mind; 1. What thickness did you use for the back rest? 2. How deep did you enter the spindle into the backrest and also the slab? What kind of drill bit/method did you use for drilling to avoid tear oout at the edges? What angle did you use for the spindles and side support, and why did you choose to match the backrest angle to majority spindle angles? Is the back rest curved along the main spindle portion, and if so is the spindle line where it enters the slab a straight line with slightly differing angles? And did you steam the back rest to achieve the curve, and does the curve increase towards the end? Any steaming referencing you could recommend would be helpful. If you could help with any of those it would be greatly appreciated and I will include some pictures of this beauty slab I am working with to sweeten the deal. It is a spalted maple slab that I milled a couple years ago and for some reason dried quite nicely, not much movement in the wood, maybe to do with the spalt. Thanks Alot317574317573317575
3. Did

[John Pond]

Thanks Alot for your reply. I have some walnut I was hoping to use about that thickness so that is good to hear. A couple questions regarding your answers.
Are you saying you laid out some painters tape on the slab and then laid out the holes on top of the tape, and then drilled through the tape? Did the tape surrounding the hole remain stuck, ie stuck nicely right at the border of the hole to protect against glue over flow? This sounds like a good idea because I was wondering how I was going to tape around all the holes.
Another question would be is cleaning up glue over flow even a problem ?
And another question would be how much of a problem would be if I were to drill deeper than intended, ( I am using a hand held drill and can only gauge depth by marking the drill bit ) and then had a space below the end of the spindle so that spindle would be glued at the right back rest height but could potentially sink if pressured? Maybe this is a non issue so long as they aren't all like that.
Also, I made a jig with a predrilled hole at angle of about 13 degrees which I clamp down and then drill through and have had fairly good result in terms of a tight fit/border at the surface but am still able to adjust the angle of the spindle a fair bit, ie. there is a bit wiggle room, but without creating a gap at the surface around the spindle. Just wondering if you found this as well, or if my mortise could be tighter and whether that is something that worth correcting, or whether the glue will take care of that. (This may be because the spindles have a reduced bottom portion, so the 1/2'' bottom portion enters and the 3/4'' portion above is meant to cover and sit on the joint/mortise, which wont work here because the reduction is cut at right angle. (But I am making a 3/4 hole and placing it deeper). I plan on cutting off the bottom to get the back height right but haven't tried that yet. (The spindles are longer than needed)
The end spindles you said you wedged underneath, did you still glue them?
Do you have a good trick for lay out? Right now I am using the predrilled block and marking both sides. Does this still work well if the block is square and your laying out a gentle curve?

And also, what was your process for fitting/gluing, did you glue the spindles into the back rest or the slab first? And did you pour glue into holes and then stick them in all at once? And hope your lay out was perfect? I was imagining that I would be able to glue the bottom in first and make sure the spacing of the spindles looked good and then mark the back rest accordingly. These are questions I am only now thinking about and I'm sure there will be more where that came from. But hopefully those questions and descriptions were understandable, I'm not too familiar with the woodworking terminology. And Thanks again for the help.

[Art Brazee]

I did layout the hole locations on the tape by direct transfer from the back. I did leave the tape on throughout the glue up. It is not a problem to drill deeper; in fact, I would recommend it. Really, your approach should be based upon the limit of your tools and what you want to accomplish. In the case of the spindles it is hard to turn them to an exact (very perfect) dimension so that each fits tightly in the mortice. However, you want the bottom of the spindles to look fantastic - it is where people will look. The top joint is unseen. You still need a good glue joint at the top, but it there where any play won't be discovered. Therefore, over drill the bottom (and top) holes, use any play to create a very tight bottom joint and allow the top to do what it's going to do. These are very small tolerances and I was quite surprised how difficult it was to get the EXACT right diameter on the spindle.

The use of a hand held drill is fine. I would take it a step further and use a brace. Mistakes are much harder to make and the results very good - maybe better. If you do use a power drill your method it good.

When I located the back on the seat I actually set it on the slab and drew the 3/4" profile on the tape. I corrected for any deviations between sides (left and right mirror), marked the holes and drilled. I used math to locate the two large spindles and dry checked the work before drilling. When I was ready to glue-up I inserted the spindles in the slab, lightly set them and starting in the middle worked in the back. Once the back was on I hammered it so it was tight and uniform. It was quite a short glue-up. If you've turned the spindles carefully this will give you good results. I used a 1/8" block with my spindle mortice drilled into it to test each spindle so I knew exactly where it would become tight. I made small adjustments with a small plane and smoothed it off with a scraper. I like to see the "working" marks in my work and don't sand so this works for me.

I did glue my wedged tenons. Dry fit everything and number your spindles. If you're hand drilling you will have some variation in diameter. You can easily overcome this by tweaking each spindle diameter. Just watch your spindle profiles and make sure nothing is too far out. You will have a spindles which won't look good with each other, but if you watch this and place them each by their "brother" no one (except a cranky woodworker) will ever know there are differences. I found turning this many spindles VERY challenging.

I did want to mention the legs. I built and used a 7 degree taper reamer with a 5 degree splay in both directions. The 7 degree taper comes from Windsor Chairs. I got online help from a chair expert in Vermont. The legs are unglued. I wish I would have used a greater splay (rake) on the legs. I think 7 or 8 degrees would be better.317873



Thanks Alot for your reply. I have some walnut I was hoping to use about that thickness so that is good to hear. A couple questions regarding your answers.
Are you saying you laid out some painters tape on the slab and then laid out the holes on top of the tape, and then drilled through the tape? Did the tape surrounding the hole remain stuck, ie stuck nicely right at the border of the hole to protect against glue over flow? This sounds like a good idea because I was wondering how I was going to tape around all the holes.
Another question would be is cleaning up glue over flow even a problem ?
And another question would be how much of a problem would be if I were to drill deeper than intended, ( I am using a hand held drill and can only gauge depth by marking the drill bit ) and then had a space below the end of the spindle so that spindle would be glued at the right back rest height but could potentially sink if pressured? Maybe this is a non issue so long as they aren't all like that.
Also, I made a jig with a predrilled hole at angle of about 13 degrees which I clamp down and then drill through and have had fairly good result in terms of a tight fit/border at the surface but am still able to adjust the angle of the spindle a fair bit, ie. there is a bit wiggle room, but without creating a gap at the surface around the spindle. Just wondering if you found this as well, or if my mortise could be tighter and whether that is something that worth correcting, or whether the glue will take care of that. (This may be because the spindles have a reduced bottom portion, so the 1/2'' bottom portion enters and the 3/4'' portion above is meant to cover and sit on the joint/mortise, which wont work here because the reduction is cut at right angle. (But I am making a 3/4 hole and placing it deeper). I plan on cutting off the bottom to get the back height right but haven't tried that yet. (The spindles are longer than needed)
The end spindles you said you wedged underneath, did you still glue them?
Do you have a good trick for lay out? Right now I am using the predrilled block and marking both sides. Does this still work well if the block is square and your laying out a gentle curve?

And also, what was your process for fitting/gluing, did you glue the spindles into the back rest or the slab first? And did you pour glue into holes and then stick them in all at once? And hope your lay out was perfect? I was imagining that I would be able to glue the bottom in first and make sure the spacing of the spindles looked good and then mark the back rest accordingly. These are questions I am only now thinking about and I'm sure there will be more where that came from. But hopefully those questions and descriptions were understandable, I'm not too familiar with the woodworking terminology. And Thanks again for the help.

[John Pond]

Thanks for the quick response. Haven't had the time to get back to the shop since I posted last but I will refer back to this when I do. When you said brace rather than clamp for the drill jig, did you mean a horizontal brace, like using the 'dogs' on a workbench? I don't have this capability yet but it sounds like a good idea as I did have the occasional spin out when the bit would catch the jig. And also, could you further explain what you meant when you said "I corrected for any deviations between sides (left and right mirror)"?
I bought my spindles from some Mennonite's in the area (who generally seem to price things reasonably), so they should be fairly uniform. But the don't have the taper on both ends, just a slight taper from on end to the other, with a 6 inch-ish uniform 3/4' base before the taper begins. So it'll just be a matter of sticking them all in the same depth so they have the same dimensions at the top. But Kudos to you for spinning them yourself.
And by hand drilling I do mean with a power drill. But I recently picked a hand drill thinking it might be good to start the holes with to avoid tearout at the top. Not that it has been a problem yet. Just wondering when the bit might begin to dull after 20 or so holes.
hHow did you contain the glue overflow during the backrest placement, ie. from running down the spindles?
The legs I took off this old garage sale table, I have sanded them but haven't cut them to size yet. They look very similar to the style of the original. Any idea how he joined them underneath? I wasn't planning on having them come right through the slab however. I have about two inches of slab thickness to work with. I was half considering sticking a threaded rod into the legs and screwing a female end underneath the slab,. I have seen this on a lot of cheaper tables but have begun to think that it might not be a great idea because people will be sitting on this not putting coffees on it. I suppose the 5 degree splay would increase the rigidity but I was hoping to keep them at a right angle. Do you know if the legs are splayed in the original? Any ideas for the leg joinery? This is my first piece of furniture so all methods/ideas are definitely appreciated. Thanks again

[Raphael Weil]

Wow love this thread, so many possibilities!