After securing the work I mark the tails making sure to also mark the waste area with an X. When cutting a long row of tails it is very easy to get disoriented and cut away the wrong material. This step of marking the waste pays dividends even for seasoned woodworkers.

As previously mentioned, cutting the tails at a height of 49" while standing on the floor is almost impossible. It's at times like this that a step ladder comes in handy. Standing on the first rung I can cut the tails quite comfortably. However, I have to reposition the ladder every couple of cuts so that I can maintain a proper stance and arm motion.

There are two features of the tail cut (when cutting them before pins) that are critical. First, the cut must be precisely perpendicular to the face of the work. Failing to do so will leave unsightly gaps, poor glue joints and weak mechanical joints. Second, you must stop the cut on the scored line marking the cut depth, and this must be true of both faces. The angle of the cut is nominally 8 degrees from vertical, but this is not critical. Neither is the width of the gap (pin width). Machine cut dovetails would all be perfectly angled and spaced, but then they would look machine cut. The human is not a machine. Hand cut dovetails are beautiful precisely because they don't look machine cut, they are all slightly different, adding to the beauty of the piece.

After cutting the near vertical cuts with the dovetail saw a fret saw is used to remove the waste. The goal is to cut as close as comfortable to, and parallel to, the scored line marking the waste depth. Leave just enough material such that the score line remains un-bruised. In order to make this cut the blade has to be "twisted" in the saw so  that when the blade is cutting parallel to the score line the saw is angled up enough to clear the tails. This is easily done with the help of a pliers.

A side note. There are two stars of the hand cut dovetail show - the dovetail saw and the chisel. The dovetail saw must have a thin kerf to make precise cuts, a rigid back to keep the blade straight and fine enough teeth to leave a clean cut. The chisel must have a finely honed micro (or secondary) bevel to give it the sharpness required to cleanly cut hard wood. It should also be made of good steel appropriately hardened and it must comfortably fit in your hands. I use a western style Lie-Nielsen dovetail saw. I have both Lie-Nielsen and Marples (now called Irwin) chisels and use them interchangeably.

The score line is scored at a depth equal to the thickness of the mating pin board. A dovetail chisel is used to clean out the waste. The chisel is placed in the score line and struck with a mallet to make a perfectly straight and vertical cut. This cut is made half way down on all waste areas. Then the board is flipped and the waste cuts finished from the other side.

The finished tails look like the photo at above left. Note a few things. Most important the walls of the tail are perpendicular to the faces and the cuts stop at the score line. Note also that I cut a little off line in a couple of place. That is not a problem because the mating pins will be marked using their mating tail and all will be happy. Compare hand cut dovetails with machine cut and see which you like esthetically. Hand cut wins - hands down - every time.

The tails are used as a mask to mark the pins. The pins are then cut with the dovetail saw and the majority of the waste is removed with the fret saw, just like the tails. The rough cut pins, shown left, are then ready for cleanup using a sharp chisel and mallet shown right. When cutting the pins it is important to leave the pencil marks while cutting on the waste side of the line as shown in the up close picture below left. This is because the tails were used as masks, therefore everything under the tails must be removed. However the pencil mark is just outside the mask and is actually part of the pin. The completed pins are shown at right below with mating tails. You will have to wait until I glue up the entire carcass to see how well the pins and tails fit together.

Hand cut dovetails are time consuming. On a project such as this there are over a hundred dovetails (a dovetail is a pin and tail combination). Each dovetail takes six cuts and two chisel cleanups plus measuring and marking. That's in excess of 800 manual operations for dovetails alone. So why cut them by hand instead of by machine and jigs? Because they are beautiful when completed.  The woodworker has much more control over the size of the pins and can vary them at will. And I believe hand cut half-blind dovetails are actually stronger than their machine and jig counterparts, though that will take a project in itself to demonstrate. In addition, as a woodworker, it is a connection with the craftsmen of the past that tugs at you, it's working in a dust free shop void of power tool noises and it's the closeness to the wood that you feel as you shape it with your hands. While I wouldn't exchange all my power tools for hand saws, I do take every practical opportunity to use hand tools to shape and join wood.

The drawer framework, shown in picture at left, consists of drawer supports (horizontal rails dovetailed into the sides), sliders (running front to back in the same plane as the supports) and dividers (sitting on the sliders to guide the drawer). Note that there are three drawers, each drawer divided from the other at both top and bottom. The dividers will guide the drawer to keep it from jamming when opened or closed.

There is a problem associated with this style of case construction. The wood grain of the carcass runs vertically on the sides and horizontally along the top, in both cases parallel to the front and back surfaces. This means the carcass will expand and contract in depth with changes in seasonal humidity. If the drawer framework were one structure glued to the sides of the carcass, either the framework could be pulled apart with this movement, or the sides could be split. Since we want this piece to last for generations we have to find a solution to this problem.

To deal with it we must first know the extent of the problem, or in this case the extent of the expansion/contraction of the sides and top. I use a CAD tool called Wood Movement Master by Kite Hill Software to help with this. If you click on the picture at right you will see calculations for white oak panels used in this application. The user enters the moisture meter he/she is using (in my case a Delmhorst 2000), type of wood, area of the country, the moisture content reading (8% in this case), type of calculation desired (Board-centric meaning board expansion/contraction), how the wood was sawn (quarter sawn in this case), number of boards in the glued up panel and the width of each board. Then press calculate and presto! The results tell me that each 5.5" board will move a total of 0.06" throughout the year, varying between 5.48" at its narrowest and 5.53" at its widest. Since there are three boards there is a total movement of 0.18" or about 3/16". If I had used flatsawn white oak the movement would have been twice that, or about 3/8".

To allow for this movement without breaking joints or splitting sides over time, I made the drawer slides "float". In the picture at left you can see that the framework is constructed with mortise and tenon joints. The front tenons are glued into their mortises and the dividers are glued to the front support and the sliders. The dividers and rear tenons are not glued to the rear support leaving them to float while riding in their mortises. In addition you will notice a gap between the rear support and the sliders (top right corner of picture) which allows for shrinkage and expansion based on the Wood Movement Master calculations. Now I can rest assured this piece will last for many generations and remain as strong as the day I constructed it.

The base face frame dresses up the front of the base and provides framing for the drawers and doors. If I were building kitchen cabinets I would use either pocket screw or biscuit construction. With pocket screws a butt joint is secured with screws that are "toe nailed" into the joint through a pocket to allow the screw to hide below the surface of the wood. These screws and pockets are applied from the backside so they are not visible. With biscuits a butt joint is secured by first machining a semicircular mortise into each piece and then using a biscuit shaped tenon to join the pieces. Neither method provides as much glue surface area as the mortise and tenon joints I used in this assembly, shown in an exploded view left, and an assembled view right. In addition, pocket screw methodology relies largely on the screws for strength, since the butt joint is end grain to edge grain, which is a notoriously weak glue joint. The biscuit is somewhat better in that the biscuit provides a face grain to face grain glue joint, however the glue area is still rather small. In my opinion biscuit and pocket screw joinery are compromises that should only be used in fine furniture when mortise and tenon joinery is not an option.

Above you can see the glue-up of the face frame to the base, proving once again the old adage "you can never have enough clamps".

To this point every part of this piece has been milled and shaped with straight, eight degree (dovetails) or ninety degree cuts. Its ogee feet are quite different. Ogee feet are smooth continuous curves on both the front and side. This piece requires six ogee feet. Two sets of two are required to form the front wrap-around feet plus two back feet. Shaping them from stock milled to overall dimensions is a five step process. The feet measure 6 1/8" tall, 1 3/4" thick and 7 1/2" long. I plan for six plus one spare to recover from a fatal mistake should I make one (and I actually did on this project). I start with stock milled to the finished height and thickness and 58" long which includes allowances for chop saw kerfs and other margin to be explained later. My goal is to produce the six ogee feet shown left below.

Step 1 is to shape the major concave curve. This is achieved by passing the stock over a table saw blade at an angle as shown right. The angle can be calculated, but it is easier to draw the desired shape on the end of the stock, start with a shallow cut and adjust as necessary. The subsequent cuts must be small and the feed rate slow since this is not a cut the table saw is designed to make. But careful attention to safety and patience will produce the desired shape. A slow feed rate also produces a smoother cut. This is important because the shaping of this cut is completed by hand sanding, which can be long and tedious if you feed the stock too quickly. It further helps to use a blade with a flat top grind since it also  enhances the cut's smoothness. Safety consciousness is crucial here because the blade is exposed. After completing the table saw shaping the surface is hand sanded starting with 120 grit and proceeding through 180 grit.

Step 2 is to cut the half-round on the front along the length of the stock. I used a table mounted router with a half-round bit. It is important to make this cut such that the thickness of the stock remains 1 3/4" when it is completed (see drawing above). This is critical because later the stock is placed shaped face down on the band saw to create the side curves.

Step 3 is to shape the convex portion of the profile. I start by cutting the stock along its length at an angle tangent to the right or outside half of the curve (see drawing above). Adjust the table saw blade angle to provide the maximum benefit to the final shaping. Next I secure the stock on the bench with clamps and shape the curve using a small block plane. My favorite plane for this task is the Lie-Nielsen 102 Low Angle Block Plane. If the stock is highly figured a higher angle block plane may be required to avoid chip out.

I finish with hand sanding proceeding from 120 grit to 180 grit. I now have a 58" long ogee shaped stock that I cut into 8" length pieces. Remember that the final foot length needs to be 7 1/2", but four of the finished feet need a forty-five degree cut to wrap-around the front corners. I like to approach this cut slowly in several incremental cuts. So I leave myself 1/2" for this purpose (this forty-five degree cut is NOT made at this time).

Step 4 begins by tracing the shape of the side profiles on the back (flat) side of each foot. I am careful to account for mirror images of the back feet and the two front corner wrap-around feet assemblies. I number the pieces to get maximum grain matching when the two sets of wrap-around feet are assembled (see the black cherry feet I crafted for a secretary desk and note the grain matching). When I place the stencil on the flat side I position it so that the narrow end of the stencil is up against an end of the stock. The other end is where the forty-five degree cut is made and I want the 1/2" margin on that end. Next I rough cut the profile on the band saw, including the half-round profile (the router can't be used for this or it will produce serious chip out). This cut should leave the pencil marks to provide guides for sanding.

Next the profile is sanded with an oscillating sander using 120 grit paper and then hand sand progressing to 180 grit. Note there are parts of the profile that can not be sanded with the oscillating sander. A file, chisel and sandpaper are used to complete those with the exception of the half-round which I leave for later.

Step 5 is to finish the shaping of the half-round. This needs to match the front half-round profile, which was shaped with a router in Step 2, very closely. Failing to shape this correctly will be highly noticeable in the final piece. I use the small block plane to eliminate most of the blade marks and for those blade marks close to the flat I use a rabbet block plane or small shoulder plane because the blade on these planes extend to the edge. A sharp chisel can pare the flat areas smooth. Finally I hand sand to 180 grit. Now I cut the forty-five degree cuts needed to join the front pair of wrap-around feet. I do this on the table saw creeping up to the final cut which will leave the foot 7 1/2" long at its longest point. The feet are now ready for glue up.

The ogee feet are attached to the carcass and then trim is applied. The shaped feet are not wrapped around the back. However, a rectangular piece of stock is dovetailed into each ogee foot to absorb side loads when the piece is slid along the floor. The pocket screws are used in lieu of clamps to hold things in place while the glue dries. They add little structural strength.

Trim runs across the grain over an 18" distance. Simply gluing it in place would eventually result in a weak joint as the sides expand and shrink with seasonal temperature and humidity changes. It may take a number of years but the end result would surely be a broken joint or split wood. Nailing the trim would solve most of the problem but they too eventually fail. A better solution is shown at right.

Here tails are screwed to the top in short sections that are spaced several inches apart. The trim is milled with what might be called two half pins which in turn creates a wedged grove allowing the trim to slide into place. Glue is applied to only the front few inches holding the side trim to the front trim. As the top expands/shrinks it is free to move sliding along the dovetails.

The hutch bottom carcass now looks like the picture at left. Things are shaping up.

At left is a completed drawer. The front is white oak and the remainder of the drawer is yellow birch. Birch is less expensive than white oak, but dense and hard. It finishes well and runs in color from pale brown to dark reddish-brown.

At right is a bottom view of the drawer. Note the hand cut dovetails. In the front are half-blind dovetails, meaning they only show on the side because of the lipped front. In the back are through dovetails, meaning they can be seen from the side and back. Notice also the tapered bottom. This allows the bottom to more easily expand/shrink with seasonal temperature and humidity changes. The drawer front is 7/8" thick and the rest of the drawer is 1/2" thick (no eighth or quarter inch plywood here). It will take a lot of beating and still last for centuries. This drawer design is ages old and difficult to improve upon.

The cupboard carcass is now glued up, shown here on its maiden voyage on the base. You can see the pins at the ends of the sides, which will receive the tails of the top, and then the remainder of the face frame will be added.

White oak is a dense hardwood. This piece is getting very heavy. I am glad the decision was made to make it in two pieces. Moving it around the shop now requires help from a friend.

Not visible in the picture, but each shelf and base top have a grove routed in them 2" from the back, so that plates and serving trays can be displayed upright.

The ship-lapped back for the base was milled from plain sawn red oak stock, which is much cheaper than quartersawn white oak. After applying the lapped back, constructing the doors and shelf, the base was sanded and finished using Waterlox Tung Oil and MinWax Wipe-On-Poly. Because these finishes build slow and thin, and oak soaks them up, I had to apply up to 10 coats. The front view is shown at left. Note the bluish reflections caused by incandescent light bouncing off white walls. The thickness of the finish causes the bluish spectrum to be visible. The grain in the doors are really visible in this view. In the side view at right the reflections are dramatically reduced  and the grain in the drawers is much more visible.

At left is the base top showing the hand cut dovetails. Look closely (double-click picture for a larger view) and you will see that each dovetail is slightly different size and taper. This actually adds character to the piece. Machined dovetails, with their every-dovetail-is-exactly-the-same look, looks sterile and bland. Also note the scribe line used by the cabinetmaker to know how deep to make the cut. This is intentionally left as another "tool mark" that is a hand-made tell-tale sign. If you look at antiques with exposed dovetails you will likely see this scribe line.

Front side drawer dovetails are almost always half-blind meaning they can only be seen from the side and not the front. They are more difficult to make because they require pin sockets (see pin sockets used in the hood of a Shaker Tall Clock). Again the scribe line is visible. The pins and tails are highlighted by the contrast of the white oak and red birch, the latter used for the drawer box. Each drawer is branded using a branding iron with my name. I am hoping that one day 200 years down the road a descendant will wonder about the craftsman who made this piece, and want to know how he is related.

The cupboard has progressed since we last showed it. The top is dovetailed in place, the v-grove ship-lapped back is on and one coat of finish applied to it, a back bottom rail has been added to receive the ship-lapped back via a tongue and grove and the trim is applied.

If you look closely you will see that the v-grove ship-lapped boards are of random width, and though not visible they are also half inch thick. This is traditional for two very practical reasons. First, half inch is strong enough for the back and it saves the additional weight of three quarter inch stock. And second, the backs were usually made from scraps left over that either had too much twist or warp to mill three quarter inch stock but would mill to half inch, and this scrap usually came in random widths. Practicality aside, I think the random widths adds to the character of the piece.

Notice how the oak darkens with the application of a clear tung oil finish, even with one coat. White oak looks rather pinkish prior to any finish but darkens nicely when applied, and the figure pops out, especially the rays, which show up only in quartersawn oak.

The doors on this project are true divided light doors. Each light is a single pane of glass and the stiles, rails, mullions and muntins are joined with cope and stick joinery. This is not a task for the feint of heart, either mentally or physically. The number of joints in a single door can seem overwhelming, and milling all parts to the exact dimensions is crucial. Any differences in mullion or muntin length can produce a door that will not go together or one that is not square. In addition there are three tenon lengths to cope with (No pun intended, but it works!). It is very important to have a repeatable procedure for milling these parts. Lastly, these parts are very small and there is ample opportunity for a serious accident, which I want to avoid at all costs.  To that end I use a setup shown at right.

You might ask why I use a shaper to drive a router bit. The answer is that I am using Lonnie Bird's Divided Light Door Set (model number 800.525.11) which is a three bit set. This set has an advantage over single bit sets in that it allows for tenons of any length. The set comes with one bit for cope cuts and another for the stick cut. I don't use the third rabbeting bit because I feel it is too dangerous with pieces this small. I set up the router table for the stick cut, the shaper for the cope cut and the table saw for the rabbet cut. I also set up my dedicated mortiser at this time because it will be used for milling mortises in muntins. Once each machine is meticulously adjusted to give matching joints I leave them undisturbed throughout the process.

The trick with this divided light set is to get the tenon length correct and to get the tenon correctly registered with the cope profile. I use a sled whose edge rides along the bit's bearing (see picture at right). But to get the correct tenon length the stock needs to hang over the edge of the sled. For this registration I use two blocks of the appropriate thickness sticky taped to my shaper fence. Then I place the edge of the sled against the blocks, push the stock up against the fence and secure it in place (shown above at left). I notch the stock to assist the cope bit from having to remove all the material - it only removes that which defines the cope shape.

The stock is 7/8" square, fairly typical for this application. But one problem is that the bearing is too far down from the stock to ride on the sled edge without the use of a spacer, which I attach with sticky tape. I also sticky tape sandpaper to the holding block to prevent the cope bit from pulling the stock into itself and destroying the squareness of the cut.

One critical check is to be absolutely certain that the sled's fence is perpendicular to the sleds edge. If not, again the cope cut will not be square.

Note that I use a sacrificial block to reduce tear out. Tear out in this instance isn't generally a problem because the subsequent stick cut will likely remove any damaged material. But I find this to be good practice anytime there is the potential for tear out. Now I can make a safe and clean cope cut with both hands on the grips and away from the bit (above left).

Safety is always a consideration in the shop, hence I perform the rabbet cuts on the table saw. I find this much safer than a router cut with such small pieces. The blade is only 5/16" high. I use a feather board and push stick to stay clear of the blade.  The short mullions are a little more dangerous than the longer muntins, but all-in-all the rabbet cuts are relatively easy compared to the sticking cut which comes next.

The stick cuts for the mullions and muntins are potentially dangerous because the stock is narrow and unstable, especially on the second cut. I use a feather board to hold the stock against the fence as it passes the router bit. This way I can remove my hand from the infeed side and place it on the outfeed side after it has passed the feather board, bypassing the bit altogether (below left).

Muntins require a mortise in the middle to allow for mullion tenons to enter from each side. This cut is performed on the dedicated mortiser prior to any of the cope, stick or rabbet cuts. It is critical that this mortise be correct or the cope and stick joints will not seat properly. In addition, they must fit snug along the sides or you risk unsightly gaps that will need to be filled, but still they almost always show upon critical inspection. The mortiser is set up and meticulously adjusted with the other equipment before any work is done on the final material. Like the other three pieces of machinery it is left undisturbed throughout the process. Finished muntins are shown below right. Notice the through mortises in the middle.

The final door will be a clean seamless matrix of cope and stick joints, the doors will register themselves square, and it will be exceptionally strong thanks to all the mortise and tenon joints involved.

The picture at left shows the finished cupboard sitting on its base top. Notice that the door has been designed such that the muntins hide the fixed shelves behind them. The plate is sitting in a grove and leans against the back. All four surfaces (base top, cupboard bottom and two fixed shelves) have this plate groove.

The v-grooved ship-lapped back shows nicely through the divided lights. The spaces separating each board will provide for expansion/contraction that accompanies seasonal temperature and humidity variations. But just as important, the vertical lines add to the overall elegance of the piece. You can visit the Gallery page to see more pictures of the finished piece.