Saw kerf question on Cosman dovetails

[Barry Vabeach]

I followed the Cosman class posts with interest, I don't have that much free time so I bought the book and DVD with the intent to work my way through step by step ( I have been doing dovetails off and on for quite some time but wanted to improve) I watched the dvd all the way through and I am having a conceptual problem with splitting the line when you mark the pins. If your mark for the pins is a few thousands into the waste area, when you split the mark, nearly half of the kerf will fall into the area that is beyond the edge of the tail - assume my saw kerf was ridiculously fat and was 1/8 inch, then wouldn't I have a 1/16 gap on each side of each pin? On the other hand, if my kerf was only a few thousands, wouldn't each pin end up being too big to fit? What got me down this line is that I use a Japanese saw, and have for 20 years so I am not converting back to Western. In the dvd and book, he doesn't mention pull stroke saws at all and so I am trying to adapt to my tools. Did he have any suggestions for using his methods with a Japanese saw or on the conceptual issue of how the size of the kerf would impact the fit of the tails.


BTW, I love his layout method with the compass - I used to do layout by eye and usually ended up with several different sized chisels because many of the sockets were a different size.

[Larry Frank]

I took one of his half day classes at a local wood supply store and went with a Japanese Saw. He quickly loaned me a different saw. I did purchase one of his new saws and it works great. However, I still want to try making dovetails with a Japanese saw.

[John Keeton]

Barry, hopefully one of the guys that took the class will weigh in, but I can give you my take on it. Please keep in mind that my educational background on this is the same as yours - book and DVDs.

Rob uses a dull knife to create a "V" which by the very nature of a dull edge, is actually held off the actual face of the tail by a few thousands or so. When you saw in that "V", half of the kerf is cutting to either side, making the cut flush with the face of the tail. At least, that is the plan as I understand it. Either way, it seems to work!!

But, it is all in the sawing. That cannot be emphasized enough, in my opinion.

[Robert Rozaieski]

I think Rob is an excellent craftsman and teacher, however, his video is slightly misleading. Cutting the joint the way he describes in the video will result in a slightly fat tail and a tight fit from the saw. Keep one thing in mind from his video though. He is using walnut and aspen. The walnut is much harder than the super soft aspen so the walnut compresses the aspen considerably, resulting in an air tight fit. He chamfers the insides of the aspen part of the joint in order to aid in getting the assembly and compression started. After planing the joint it's perfect. Using two woods of such different hardness leaves a lot more room for error.

When using two boards of the same species, the fit needs to be a little more precise to avoid splitting the pin board. When marking the pins from the tails for example, I saw fully on the waste side of the line and pretty much leave the entire line in order to avoid a sloppy fit. However, saw too far from the line and the fit will be too tight and risk splitting the pin board. I do agree with Rob that going from the saw straight to assembly is best as any time I've needed to pare a dovetail joint, I usually make it worse.

[Barry Vabeach]

Guys, thanks for the input. I caught the part about a "dull" marking knife, though I haven't gotten that far yet. Thought it might make more sense to practice the cutting to the line for the tails a bunch ( parallel lines within a 1/4 inch of each other halfway across the board in one direction, the rest are in the other direction) before I completed a joint.

[rob cosman]

Hi Barry, read your question and thought I would offer a few comments. I do dull my knife a bit, this makes the line easier to see and makes the knife a little less likely to follow the grain. Don’t push the knife too deep into the endgrain of the pin board, heavy enough to see but too deep and you will loose the precision of your transfer. The knife leaves somewhat of a "V" shape in the pin board, you need to split this with the saw so that half of the "V" is still there when you’re done. One of the recent comments is a bit backwards, when putting a soft wood into a hard wood you are better off leaving all of the "V", this will compress and make for a tight fit, with two hardwoods you have to get much closer to splitting the line, if you don’t the pressure may split one or the other. The purpose of chamfering the inside of the tail is primarily to prevent the glue from getting pushed off and starving the joint. The chamfer channels the glue.

It is tough to find a good knife, most of those offered by tool companies are way off. I will have one available in the next few months, it has been under development for over a year. In the mean time a Swiss Army knife is a good tool for the job. Too thick to fit in the kerf of most saws but it can be ground thinner. The shape of the blade is a wedge from top to bottom so the transfer of tail to pin is accurate.

On a final note, no matter what wood I use you just can’t make everyone happy! I try to imitate what would be used in actual work. Since most of the dovetails you will cut are in drawers I use a fancy dark wood for the front and a white secondary wood for the side. I would love to cut my demo dovetails in Ebony and Holly but to date no one has offered to pay for the wood. When you do thirty or so in a weekend it gets a bit expensive. Walnut and Aspen or poplar contrast nicely and if you’re going to take the time to cut the joint then make sure it gets noticed! Hope this helps! Cheers
Rob

[Barry Vabeach]

Rob, thanks for the post. Again, loved the DVD and the book. I will be looking forward to seeing your knife.

[John Keeton]

It is so refreshing to get responses from people like Rob, Shiraz, Rob Lee, and others in the industry. THAT is just one of the things that makes SMC such a great place!

[Robert Rozaieski]

One of the recent comments is a bit backwards, when putting a soft wood into a hard wood you are better off leaving all of the "V", this will compress and make for a tight fit, with two hardwoods you have to get much closer to splitting the line, if you don’t the pressure may split one or the other.
Thanks Rob, that's what I meant to say but it didn't come out in writing the way I was thinking about it in my head :o.


On a final note, no matter what wood I use you just can’t make everyone happy! I try to imitate what would be used in actual work. Since most of the dovetails you will cut are in drawers I use a fancy dark wood for the front and a white secondary wood for the side. I would love to cut my demo dovetails in Ebony and Holly but to date no one has offered to pay for the wood. When you do thirty or so in a weekend it gets a bit expensive.
Sorry if I sounded like I was criticizing your wood selection. I certainly wasn't doing so and never would. Trust me, I can't do what you do. My work is no where near good enough to be critical of anyone :D. My intent was just to make the point that it needs to be done slightly different when using a hardwood & softwood and when using two hardwoods. I wasn't sure what the OPs intent was (i.e. drawers or case) so I just wanted to bring up the minor difference.

I'm glad you could chime in personally. It's always nice to get clarification straight from the horses mouth when folks have questions like these. Thanks!

[Mike Brady]

I know we are commenting primarily about Rob's methods here, regarding the marking methods, but the person who taught me (Jeff Miller) uses an awl to mark and transfer the layout instead of a (dulled) knife. He sometimes runs a dull pencil in the awl marks, which gives you two closely parallel pencil lines: one to cut and one to leave. The awl is easier to articulate when tracing the pins onto the tail board, also.

[Ben Silverman]

If you want to take the guesswork out of marking out the pins from the tails, take a look at the Glen-Drake system (glen-drake.com/). Kevin uses a feeler gage to offset the tail board from the pin board for marking. The offset is equal to the saw kerf plus the user determined tolerance. He then marks the cuts to the waste side using his kerf starter, a scraper, having the same thickness as the saw kerf. The saw starts easily and precisely in the scraper kerf. The combination of the feeler gage and the kerf starter takes the guess work out of layout. Needless to say, it does not take the skill out of cutting in two planes. Ben

[Danny Burns]

It is so refreshing to get responses from people like Rob, Shiraz, Rob Lee, and others in the industry. THAT is just one of the things that makes SMC such a great place!

I'll second that! :rolleyes:

Thanks Rob for your response! :D

[David Keller NC]

Folks - One comment about using Rob's method to set out the pins/tails with a divider. A few other dovetail gurus have sort of back-handedly "poo-pooed" the precision that the divider method, stating that it slows productivity down and isn't necessary because one side of the joint is used to mark out the other. This is absolutely true.

However, there is a very, very good reason and advantage to using the method that Rob teaches for precision layout of tails/pins, and that's error recovery. You're unlikely to do this on a drawer, but I'll describe a situation I ran into when making an 18th century mahogany bookcase. The bottom and top boards were mahogany, as were the case sides, and they are dovetailed together at the corners. Like most bookcases, the back edge of all 4 pieces are rabbeted to receive a nailed series of backboards.

Well, I really blew it here - I cut all 8 joints for the outer case (4 sets of pins/tails) - the tails are on the top/bottom boards, the pins on the long case sides. What I messed up on was clearly marking the corresponding pairs of pins/tails, so when I cut the rabbets for the case back, I accidentally cut the rabbet on the front side of the case top.

As you might imagine, I wasn't too happy about this, and the material was very expensive figured mahogany on the case sides. I had a few choices - one was to cut off the pins/tails set on the top of the case sides and top, and re-cut these joints with the top flipped around so that the rabbet was on the back side again. Another choice would have been to cut down the width of the sides, top and bottom and re-cut all 4 rabbets on the correct back side (I left fairly large half-pins on the outer two edges of all 4 case sides, so I could do this and get away with it without messing up the dovetails). Finally, I could mill another board of mahgany fot the case top, and set aside the original top as a reminder to not screw up numbering the pairs of pins/tails.

The last option was out - I didn't want a board with that big of a mistake setting in my shop and taunting me every time I was there. :) Re-cutting the joints would've been laborious, and cutting the pins on the case sides took some doing. Because of the length of the sides, the only way to cut these was to stand on the bench to get to the height necessary to have a proper sawing stance to cut the pins. I got away with it the first time, but wasn't sure I would the 2nd. Finally, I'd carefully set up the width of the case to fit some of my rare and expensive books, so I'd have had to find a new home for these if I'd "redesigned on the fly".

So the solution? I tried test-fitting the joints as cut so that the top board was reveresed from the way I'd transferred the tails to the pins on the case sides. I know Rob doesn't like to do this (test fit pins/tails), but when you've this much at stake, splitting one corner because the pins/tails are too tight is not an option. Guess what? Because I'd precisely laid out the tails to be symmetrical across the width of the board with a divider, used a dovetail marker to set the tail's slope, and sawn carefully, flipping the top around so that the rabbet was on the correct side did not destroy the fit of the two upper corners, irrespective of whether the pins inserted into the tail sockets were actually marked out from that series of tails or not.

To say that I lucked out on error recovery is a serious understatement, and there's absolutely no way on this green earth that I could've gotten away with this if I'd simply eye-balled the tail spacing when I cut the first side of the joint. After this experience, I always mark out tails with a set of dividers - regardless of the small amount of extra time that it takes.

[Barry Vabeach]

David, suprised that you were that consistent that it fit both sides. I had heard his method downplayed as too mechanical and that they should be laid out by eye, but when I cut 4 sockets and need 3 chisels because one socket is 1/8 smaller, and another is 1/8 wider than the other 2, it seems any time saved in marking out is quickly lost later in the process.

[David Keller NC]

I'll admit that I had to "persuade" the joint on one side to go together with a rubber dead-blow mallet, so I wasn't that precise. But I rather doubt that the joint would've gone together at all, even with additional paring, if I'd eye-balled the pin/tail layout.

And I wouldn't at all be surprised if furniture historians/scholars came to the conclusion that urban 18th century cabinetmakers used the divider method. I haven't been able to examine a lot of the furniture from the period in that kind of detail, but I do have one very early 19th century Sheraton-style dropleaf table made in New England that I lucked into 10 years ago or so. The single drawer in this table is absolutely without doubt made completely by hand, and the dovetails on both sides of the drawer front are extraordinarily precisely laid out - I can't measure the difference in tail widths, which would mean they're accurately the same width down to about 1/32nd of an inch.

[Joel Moskowitz]

And I wouldn't at all be surprised if furniture historians/scholars came to the conclusion that urban 18th century cabinetmakers used the divider method. I haven't been able to examine a lot of the furniture from the period in that kind of detail, but I do have one very early 19th century Sheraton-style dropleaf table made in New England that I lucked into 10 years ago or so. The single drawer in this table is absolutely without doubt made completely by hand, and the dovetails on both sides of the drawer front are extraordinarily precisely laid out - I can't measure the difference in tail widths, which would mean they're accurately the same width down to about 1/32nd of an inch.

I think the chances of this are zero. Laying out stuff by eyeball can be extraordinarily symmetric and when you take into account tons of practice and a quest for speed there would be no reason to use dividers. See how Frank Klausz does it. For a modern person who doesn't have the practice, and really wants a specific layout the divider technique is one way prefectly reasonable way of doing it.

[Billy Chambless]

I think the chances of this are zero. Laying out stuff by eyeball can be extraordinarily symmetric and when you take into account tons of practice and a quest for speed there would be no reason to use dividers. See how Frank Klausz does it. For a modern person who doesn't have the practice, and really wants a specific layout the divider technique is one way prefectly reasonable way of doing it.


You just blew a chance to sell a bunch of dividers!

;)

[Joel Moskowitz]

You just blew a chance to sell a bunch of dividers!

;)

:) ha ha ha

[Jan Bianchi]

I've taken a couple of Rob's classes, and really enjoyed them, but I think the V, or splitting the line, is a little confusing. Everytime I cut down the center of the V, I take too much from the pin side so the fit is loose. Instead of a V, I think of the line I'm going to cut as two long, contiguous, vertical rectangles. Don't cut down the center between the two rectangles. Instead cut completely down the rectangle on the waste side of the line. Leave the center line between the rectangles and the pin side rectangle in tact. If you think of it this way you should still be able to see half the V on the pin side of the line after you've made your cut. My recollection is that Rob drew this on the board in class, but he doesn't explain it that way in the DVD. Anyway that works for me.

[Barry Vabeach]

Jan, that makes a little more sense to me in terms of layout. The thing that stood out to me is that in theory, you can get tight fitting dovetails no matter the size of the kerf of your saw, OTOH, if you went with the split the line approach, the wider the kerf of your saw, the wider the on fitting the pin.