Old Oud - New Project

Following the picture posted by Peyman, this is an appropriate moment to consider another design detail - the tuning pegs. As can be seen from this image, the peg heads appear to circular in profile unlike the violin peg shape often found on ouds today.
The pegs on the oud engraving also seem to have circular peg heads as do the pegs on, for example, on the familiar oud like instrument depicted in the 13th C. Cantigas de Santa Maria. Both of these examples also show the pegs to have long shanks. This way of representing the pegs may, of course, just be an artistic convention - simply because they are easier to draw that way perhaps.
Nevertheless, the pegs on my old Egyptian oud are circular in profile so this is the style of peg that I will be making for this project but with somewhat shorter shanks - long shanked pegs being weaker.
The pegs will be made from well seasoned Pear tree wood that I cut into billets for making pegs about 30 years old. The tree was given to me by a friend and neighbour who thought that I might put the wood to good use. As both he and his wife are - sadly - now deceased, the use of the wood for this project will be a small memorial to them. I have about 70 billets in stock so there is plenty of choice.
The bridge will also be made from this material.
The wood is very fine grained, working nicely under the plane, so should be good for turning also. I shall measure a sample later to determine the specific gravity.

Meanwhile - back to the work bench!
On inspection, there is a problem with two of the glued panels. I was unable to work quickly enough in coating the glue joint and positioning the panels before the glue in the joint started to gel. This prevented the joint surfaces coming tightly together leaving a very slight but visible (and, therefore, in my judgement, unacceptable) gap in the joint.

The method of gluing the panels has, therefore, been modified to overcome this problem - based on the method generally used for gluing sound board to bowl. This involves reheating the glue so that it melts in the joint - using a hot iron - allowing the joint surfaces to come tightly together under the clamping pressure.
An iron heated over an open flame could be used for this purpose but I am using a small electrically heated, thermostatically controlled, iron for convenience. These irons may be purchased from hobby shops (used for applying plastic film to model aircraft?) at a reasonable price. They can also be purchased from luthier supply houses at double the cost.
When the glued joint surfaces are pressed together on the work surface, the hot iron is simply moved over the joint until the gelled glue is seen to re-melt and the joint fully close. The seam is then clamped to the flat work surface using a straight wooden batten and two 'G' clamps as previously described.
The setting of the iron thermostat is in the mid position. The iron has a 'Teflon' sole so that any surplus glue will not stick to it.
My glue pot is simply a small, screw top glass jar set in a pan of boiling water to melt the glue. The hot glue is diluted with water to the correct consistency - running off the glue brush in a steady stream. After use, the jar is sealed and left in a refrigerator for use the following day. This should be good for about a week after which time it will be thrown away and a fresh batch of glue will be made up.
The good old fashioned cast iron glue pots that can still be seen in some professional luthier workshops (I have a couple myself) are not required for occasional, small production, work. A glass jar is cleaner, cheaper and holds small measured quantities of glue so is very economical in use.

To follow the complete analysis and arguments that has led to the decision to move ahead this project, you will need to wade through "Analysis of an Early Oud Woodcut" (Advice, Tips and Questions Forum" and also check out "Below Bridge Bars - a Missing Link?" (Ouds Projects Forum).
However, in order to briefly summarise the situation I will reference the attached geometrical diagram used to analyse the barring (a speculative proposal) of the 12thC/13thC engraving of an oud compared to the barring geometry of a lute given by Arnault de Zwolle circa 1450.
The Arnault drawing is 'dual purpose' showing the barring geometry as well as the bulkhead placement of a mold required to build the lute. One striking feature about this lute is an 'unusually', massive end block below the bridge with no other bars. The oud does not have an end block but a 'lute like' end plate of only a few mm in depthas well as a first bar.

The front edge of the bridge in both the oud engraving and the Arnault lute is 1/6 of the body length which is the classic lute bridge position - confirmed later in the early 17th C by the great French theoretician Marin Mersenne. (see the reference scale alongside the oud diagram for comparison) as well as on surviving lutes. Lutes (judging from surviving lutes) did not have a first bar below the bridge but instead had a kind of fan bracing arrangement with a "bass bar" approximately aligned with the first bar position on an oud. Either way, the objective would seem to be, for lute or oud, to significantly stiffen the sound board area below the bridge.
Note that the bridge length in the oud engraving is way off scale and will be much shorter (and narrower) than represented as it must carry only 5 double courses. There is, in fact, physically enough space for the first bar below the bridge according to this geometry (the bars shown as dashed lines in the geometry will be incorporated)
I do not anticipate any problem with this barring arrangement but, of course, will have to build the oud to find out just how it all works - if at all!

Not at all!
I have built copies of surviving lutes with the classic 1/6 bridge placement (including the Arnault lute) as well as lutes with alternative bridge locations. The European iconography of the 15th C and 16th C shows many examples of bridge placement on lutes at variance with this rule - although it is risky to depend too much on evidence presented by paintings when attempting to recreate early instruments.
One fragment of an early lute belly survives with clear markings of an original low bridge position. This is from a lute by the renowned maker Laux Maler - cat. No. M154 in the Nurnburg Germanisches Nationalmuseum. Other bowl fragments by Maler survive in other European museums - providing sufficient evidence from which to make a replica.
Here is an image of the belly of a successful reconstruction that I made of the Maler lute in 1979. Note the low bridge position.There is no bracing below the bridge either.

Off topic - so just for general information and interest.

The first of the batch of four panel sound board blanks is glued up, dried and ready to go.
The first step is to level the blank evenly on both sides to remove any slight 'step' at the panel joints. The blank is currently about 2.5 mm to 3 mm thick. Leveling is done using a cabinet scraper (I do not trust the smoothing plane as it may cause excessive tear out of the grain). The scraper is worked in all directions across the surface of the blank. To highlight any high spots in the surface, a toothing plane is also used. (The blade on this kind of plane produces fine scratches on the surface - originally used for preparing the glued surface of panels prior to veneering). The scratch marks are then removed with the cabinet scraper to produce a smooth finish overall.

The next step will be to cut out the blank to the required profile (cut well oversize by about 12 to 15 mm) prior to final thicknessing to around 1.5 to 2 mm (1 mm thick in the sound hole areas to facilitate cutting in of the rosettes).

In the meantime, work has progressed with preparation of the rib blanks. (It helps to switch between tasks to relieve any work tedium - but not the most efficient way to go). Two partial rib sets are to be used - alternating each rib of the bowl in sequence and matching grain pattern.
All of the rib blanks have been prepared on one side (the inside face) by planing and final smoothing with a cabinet scraper. The current thickness of about 2mm is well oversize so will be reduced to about 1.5 mm final thickness by smoothing the outside face with a cabinet scraper.

The rib blanks have been cut oversize to equal length (less material to remove during the final stages of thicknessing).
To ensure that the ribs are used in sequence during assembly of the bowl the ends of each stack of blanks has been marked with red ink.

The sound would be 6cm shorter te he he sorry fellas could'nt resist

Thanks charlie oud - I was wondering!

The second sound board blank was glued up and completed today. The original pair of panels - assembled without using a hot iron to remelt the glue - are not satisfactory as the glue has not fully penetrated the joint surfaces. They have, therefore, been taken apart by first wiping the joints with a damp rag and then using the hot iron to melt the glue and separate the joint surfaces. Another big advantage of using hot hide glue - it is easily reversible. The panels have now been re-jointed in preparation for making sound board blank #3.

This batch of glue has been in use for a week so has now been discarded and a fresh batch made up. Before throwing it out, the heated glue was diluted with hot water to the consistency of a thin varnish and brushed over the neck block surface. This 'glue size' is to seal the grain of the neck block preventing the end grain of the wood from absorbing too much glue during assemby of the ribs to the neck block (which could result in a weakened joint).

The second neck block is from another project (see 'Old Project - New Lute' on the Forum). Thought that I would use the opportunity to 'size' that at the same time.

The first of the sound board blanks to be completed has now been smoothed on both sides to remove any surface glue and reveal the grain figuring and any faults. A pattern of half the sound board profile - cut from thick card - is used to determine the best part of the blank for the finished soundboard.

The blank will be cut to the required profile - oversize by about 12 mm - for trimming later. A good method to use for marking out an oversized profile is the old trick of using a metal washer - rolled around the outside edge of the card pattern, like a wheel - with a pencil held in the centre to trace the profile. (This style of washer - with the relatively small centre hole - was obtained from a local auto parts dealer).

As four (or five) sound board blanks are being prepared, the best of the bunch will be selected and the rest then kept in stock (untrimmed) for future projects.

For sound board blank #4, the method has been modified slightly to improve the procedure for gluing the joints. This is the final blank of this batch.

By the time both edges of the joint have been coated with hot glue and pressed in place (under pressure), the glue has gelled sufficiently to prevent the joint faces coming fully together. The glue must, therefore, be remelted with a hot iron. This time - instead of applying the hot iron directly to the joint - a strip of cooking paper was first laid over the joint followed by the iron. This procedure is less messy and works well.
After the joint surfaces have moved into position they are held in place, flat on the work surface - as before - with a wooden batten clamped over the joint.
The cooking paper is 'Baker's Mate non stick parchment paper'. It is for use in the oven or microwave - 'greaseproof', waterproof, temperature resistant to 215 C (420 F), strong and reusable. Good stuff for this application.
There are likely other similar brands available on the market. Do not used waxed paper!

The precision of your work is REALLY inspiring.
I am going to use some of your methods the next time I make an oud -
Thanks for sharing all of this!

The rib blanks have now been planed and finished to a thickness of about 1.5 mm with a cabinet scraper. Using the rib template to evaluate the best part of the rib blanks, the rib centre position has been marked on each blank for reference during bending.
Nothing special about the hot bending technique - just using a heavy duty propane heated bending iron - dry heat no moisture. (Note - I do not use a kind of special single handed technique - one hand must hold the camera!). The modified jig is used as a guide to the correct rib profile. The rib centre mark is used as a reference to ensure matching grain pattern between ribs.

Each completed hot bent rib blank has been set in a simply made carrier to maintain the rib profile and reduce any spring back. Each rib will eventually have to go back on the bending iron for adjustment to the correct profile prior to fitting and gluing.

All of the rib blanks have now been hot bent to the required profile.
So, for a change of pace, work on the selected #4 sound board blank has started. Both surfaces must first be planed level - a preliminary operation prior to final thicknessing. To avoid tear out of the grain, the blade of the smoothing plane has been freshly sharpened and set for the finest cut possible. The plane is worked in all directions - crisscrossing diagonally across the grain - until all high spots and surface glue have been removed. The plane marks still remaining will be removed during the final finishing stages. The rough finished sound board blank is now just over 2 mm in thickness.
The sound board profile has been marked oversize and will be cut out on a bandsaw ready for the final finishing process.

Although the soundboard is cut well oversize it was decided not to use the bandsaw to avoid any risk of tearing and splintering of the edges - easily done when cutting Spruce across the grain with a coarse saw blade.
Instead, the soundboard was cut out by hand - first using a razor saw to trim off most of the waste material. The profile was then scored along the marked pencil line with several light passes of a sharp craft knife - working with the grain direction - to about a millimeter or so depth of cut. The final cut was made using a very fine jeweler's saw blade mounted in a fretsaw frame - sawing along the knife cut. This operation was fairly quickly and successfully accomplished without any splintering. No need for expensive power tools!

Jeweler's saw blades are graded according to fineness of cut, in a range of 20 sizes, from 8/0 the finest to 14. Not sure about the size of this blade (probably about 4/0) but the teeth are so small that they can only be seen with a magnifying glass.

The sound board surfaces were then inspected and the 'best' side chosen for outside face of the sound board. This face was finished using a cabinet scraper until flat and smooth. The surface was checked as work progressed by feeling for high spots and irregularities and by eye - looking across the surface at a low angle under incident light. Human touch can be a very sensitive means of detecting surface variations (of about 0.001 inch or 0.025 mm).
This face will now be left untouched and the final thicknessing will be achieved by removing material from the other side (the underside) of the soundboard.

The 'underside' of the sound board will first be made flat and smooth with scrapers and of uniform thickness all over - target 2mm thick initially. This will then be gradually reduced in thickness according to how stiff the sound board feels as work proceeds.

At this stage the toothing plane will not be used for levelling although here, for information, is an example of the underside of the sound board of an old Egyptian oud - left 'as is' after using a toothing plane. Who says that the underside of an oud sound board should be made silky smooth?!
What was the intention of the luthier here? To produce a better gluing surface for the braces? Or, perhaps, a subtle refinement modifying the sound board to be more flexible across its width - presumably with acoustical benefits? Or did he just make a mistake and remove too much material with the toothing plane and let it go at that?

Anyway, the sound board underside on this project will be finished smooth and uniform - at least for first time around!

(For those who may be wondering, the restoration of the Egyptian oud, subject of the attached images, is still pending and will resurface on the Forum at a later date)

Another small interlude.
I plan to measure the final thickesses (the thickness will vary a little from area to area) across the entire surface of the finished soundboard. This is mainly for record purposes and is not essential.
The measurement will be made with a luthiers caliper. 'Luthiers Mercantile International' at http://www.lmii.com sell calipers of this kind - nice tools but costing around $150 US. So I decided to make my own - not as fancy but one that will do the job.
The heart of this measuring device is a dial (or deflection) gauge made in China costing around $15 US or less (but good quality nevertheless) - available from most engineering tool suppliers as well as Lee Valley Tools. They come in various dial sizes and travel as well as being graduated in Imperial or metric graduations. (This gauge, for example, has a 2inch dial, 1 inch travel and graduations in thousands of an inch).
The frame of the calipers has been made from 7/8 inch (22 mm) thick straight grained pine - the two arms being joined together with a spacer block and open ended mortice and tenon joints. This type of joint is easy to accurately cut on a router table. The image shows the spacer block prior to trimming the tenons and finishing. The throat of the caliper is 9 inches (23 cm) so will cover sound boards up to 18 inches wide. The dimensions can be varied according to personal requirements and whim. The design principle here is "if it looks right it is right"!
The dial gauge is clamped securely in place on the upper arm with two machine screws. A saw cut through the gauge mounting hole provides some local flexibility for clamping.

The frame has been assembled using brand "Gorilla" polyurethane glue. No particular reason other than I wanted to try it. Any good cabinetmaker's wood working glue would do just as well.Once the glue has cured, the frame will be trimmed and finished.

The calipers will be clamped in a vice in use allowing both hands to be free for manipulating the sound board.

Thanks Samir
Why not go 'mobile' and do the forum on your lunch break?! Apparently
A.T &T are touting a mini laptop (Acer Netbook) for only 50 bucks U.S. - the catch? - you must sign up for a two year contract ($1500 US ?).

The calipers were completed this morning by first trimming and finishing the frame - rounding off sharp corners etc. The final step is to fit an 'anvil' to provide a raised, point contact surface with the plunger of the dial gauge. The dial gauge is first mounted and clamped in place and the plunger carefully pressed down to make a slight indentation in the lower arm. This indentation in the wood is the exact position for the centre line of the anvil. The anvil may be made from metal or wood and should, ideally, be dome shaped. Many items of hardware might be adapted to serve as an anvil - dome headed rivets, small carriage bolts, ball bearings (or even a piece of wooden dowel rounded into a dome) etc - all available, off the shelf, from hardware stores. Here I have used a brass, dome headed, upholstery nail together with a hard rubber cabinet door 'bumper' that I happened to have in stock. The nail/bumper combination were hammered into place by removing the dial gauge and using a wooden dowel as a punch - the dowel diameter being a close sliding fit in the dial gauge mounting hole as a guide.

Before use of the calipers, the dial gauge must be adjusted to read 'zero' when the plunger tip is in contact with the anvil. A rough adjustment can be made by positioning the dial gauge in the frame before clamping it in place. Fine adjustment is then made by rotating the scale of the gauge to the zero mark.
In use, the caliper is mounted in a bench vise and the sound board manipulated with both hands. The sound board should be held level to obtain the most accurate readings.
The plunger of the dial gauge is spring loaded so this provides the light pressure necessary for an accurate reading but is an insufficient load to deflect the frame and cause a false measurement.

All of my accurate measuring tools, calipers, dial gauges, micrometers etc are in standard Imperial units of thousands of an inch because those are the units that I have always used for working metals - lathe work etc. However, I prefer to use Metric measurements for luthier work. So - when using an Imperial standard dial gauge for instrument making - all that is necessary is to make up a handy chart converting the Imperial readings to metric. This chart needs only be in steps of 0.1 mm.
A simpler solution, of course, would be to use a Metric dial gauge.

Work on final thicknessing of the soundboard - using scraper blades - has progressed to completion. Thickness is 1.5 mm in the central part of the soundboard increasing to 1.8 mm around the edges.
As the sound holes are to be 'cut in' - like a lute - the area covered by each sound hole must be reduced further in thickness to about 1 mm (0.040 inch). At present only two sound holes are to be cut - as represented by the oud engraving - but a third sound hole may be added later. Using a scraper blade, and working around the sound hole centres pierced in the sound board, the thickness was carefully reduced - regularly plotting the reduction in thickness as work progressed. The thickness calipers proved very useful for this operation.
An assessment of the relative sound board thickness variations can be determined by holding the sound board up to a strong back light.

Thanks Jameel. The Chinese made precision measurement tools can be great value today.

The next stage in cutting the sound holes is to glue a paper pattern of the rosette design to the underside of the sound board. A traditional 16th C lute rosette design has been selected as this clearly has Arabic geometric design origins - as well as the six pointed star motif represented in the oud engraving. In order to match the sound hole diameter it was necessary to reduce the scale of the original lute pattern to 78%. This was done on an ink jet printer using 20 lb paper - the ink is water soluble so the paper pattern was waterproofed with a coating of orange shellac. The paper pattern was glued to the sound board with thin hot hide glue - applied to the sound board not the paper (which would curl up and be difficult to handle). To ensure that each pattern was fully glued, they were ironed over with a warm iron.
To reinforce and harden the front face of the rosette in preparation for the cutting operation, the soundboard over the sound hole area was coated with shellac. The shellac will be removed later after completion of the rosettes.

The rosette cutting - or piercing - work is done with a fine chisel cutting directly through the pattern on the back of the sound board with the front face of the sound board supported on a hard, flat, smooth surface. A piece of 3mm thick tempered 'hardboard' makes an ideal surface.
The chisel has been made from a broken hacksaw blade - glued into a wooden handle with epoxy cement - and ground into a narrow profile as shown in the images. The width is just under 2 mm. The cutting edge is honed to a fine bevel and is curved along its length. The cutting action is to stab downwards through the sound board thickness at each corner of the pattern and then to extend the cut by rotating the chisel backward. Slicing cuts (using a standard scalpel or craft knife) are not practical due to the hard/soft nature of the grain of the Spruce - making the knife impossible to control.
Cutting starts at the centre of the rosette working progressively outwards. Each piece, when correctly cut, will 'pop' out (from front to back) of its own accord - it must not be forced to avoid the risk of breakage of the delicate wood tracery. I do not bother to try to remove little 'whiskers' of wood left behind (the rosette is very fragile and easily damaged) preferring to leave everything as it comes 'off the chisel' - 'warts and all'!
Every few cuts, the chisel should be honed on a leather strop (dressed with Lee Valley 'green' compound) to keep the cutting edge razor sharp - essential for clean cuts and to avoid breakage of the wood.
It takes me about two hours to pierce a sound hole rosette of this diameter (70 mm). It helps to take a break from the task every so often! Boring but satisfying in the end!

Here is a pierced but incompleted rosette as viewed from the underside of the sound board as well as on the front face.
The next step is to emboss the front face of the rosette to emphasise the interlacing elements of the pattern.

Thanks Samir.
As 'cut in' sound holes must be completed before a sound board is braced, a final decision must be made about the sound hole configuration - two sound holes or three? The oud engraving represents only two but perhaps a third, centrally placed sound hole was omitted by the engraver for clarity?

The attached low resolution image shows an unidentified picture of an early oud (10th C Persian?). Note the two large sound holes placed low down near the bridge as well as a smaller sound hole located near the fingerboard. (Note also the sickle shaped peg box and soundboard edge banding).
For comparison, the other image shows an early European (15th C) fretted lute with a single large sound hole placed, low down, near the bridge, with a smaller sound hole located closer to the fingerboard.

Edited 27 November 2009 to include higher resolution image of the oud.

Here is another example of two early 15th C lutes (Florentine) each with a large sound hole placed near the bridge and a small sound hole located higher towards the finger board. Note also the lack of frets, the five courses, the ribbed construction of the bowls and sloping shoulder of the neck at the neck joint. Definitely oud like features! Note also the Arabic geometry of the large 'cut in' sound holes and the similar geometry of the reduced diameter sound hole.

I would speculate, from this limited information, that some early oud/lutes had either two sound holes or a single sound hole located near the bridge but with a smaller diameter sound hole located higher up on the sound board - the equivalent total sound hole areas being the same in each case.

So, rather than adding a third large diameter central sound hole (as used in creating the geometry of the oud soundboard) which would have significantly reduced the sound board area, I have chosen to add a small diameter sound hole fitted between braces five and six.
It so happens that the total area of the three sound holes is within 7% of the area of the single large sound hole taken from the oud sound board geometry.

So the little sound hole is now 'cut in'. No turning back now!

Thanks DaveH. There are at least 8 historical possibilities for sound hole configuration in ouds/lutes plus their variations in placement on the sound board. However, a large central sound hole together with two medium sized sound holes does not appear to be one of them and it does not 'feel right' for this instrument either (excessive sound hole area).
I hope to be able to make some comparisons with this arrangement by temporarily blocking off the small sound hole with a thin aluminium plate, taped in place, just to see if that makes any significant difference.

Here, for information, is the image of a Tunisian four course fretted oud (oud arbi?) previously posted on the forum that retains the two large/one small, sound hole configuration but positioned as a group high up on the soundboard. The rosettes on these instruments were often 'cut in' apparently.
This close grouped triple rosette arrangement was also to be found on some of the larger lutes of the 16th/17th C.

On the subject of historical sound hole configuration I shall next post a brief research article on the subject - for what it is worth - that I wrote for FoMRHI in April 1979 and which is listed in the bibliography of Robert Lundberg's book "Historical Lute Construction" so it may still be of some general interest.

The final step in cutting the rosette (shamsa) is to emboss the front face to create an illusion of the interlacing pattern. Traditionally this was done by lute makers by means of chip carving using small chisels and knives. Wood engraving or block cutting chisels cannot be used for this work as the rosette is too thin and delicate. I use a scalpel to slice shallow 'V' grooves to about 0.5 mm deep in the tracery of the rosette (not much material is left after this - a delicate and time consuming operation). However, as this is an experimental project and as my eyes and steadiness of hand are not quite what they once used to be, I have decided to undertake the embossing work not by carving but by wood burning using pyrographic techniques - a hot blade being used to scorch grooves in the wood to create the embossed pattern. Pyrography is a very ancient technique - at least 3000 years old - for producing incised patterns in wood or other materials. It is a technique that was used to decorate the ancient harps of Ireland and Scotland but to my knowledge has never been used historically in the creation of lute rosettes. So here is a first! The black, charred wood lines should create an interesting contrast - hopefully.
First of all, the incised lines are marked on the rosette in soft pencil as a guide. The pattern is complex so it is easy to make a mistake and mistakes are not an option at this stage!

The pyrography 'pen' that I am using is a low cost tool purchased from a local hardware store - essentially just a low wattage (30 Watt) electrical soldering iron with a chisel shaped tip. Better pyrography units with temperature control and a variety of tip shapes - used by bird carvers - are quite expensive but are likely more controllable and would do a more precise job.
A steady hand and very little pressure on the pen - just a light touch - is required as it can be easy to burn through the wood especially in the direction of the grain.
The end result is not quite as neat and precise as I would have liked but is sufficient to prove this technique as viable - and a lot faster than chip carving. I just went ahead and did the work but a bit of practice before hand might have been of benefit!

As an after thought, the tip of the pyrographic pen has been reduced to about half its original length by filing. This allows the pen to work more precisely. Too late for work on the rosettes - now essentially completed - but useful for final 'touch up' of the details.

Et voila! - the completed (but unbraced) soundboard.
We have been experiencing unusually dry Spring conditions of low relative humidity so this is a potential 'window of opportunity' to complete the bracing of the soundboard. However, the rain clouds have already moved in so this may no longer be a possibility - in which case, barring of the soundboard will have to be postponed until dryer conditions prevail towards the end of this year. No problem - lots more work to be done.

With cool outside temperatures and the wood stove in operation, relative humidity in the kitchen dropped to 50% overnight so the sound board was braced this morning. This may be the last chance before hot, wet weather starts to move in this weekend.

The braces have all been cut and planed to an equal height and thickness of 15 mm X 4 mm but will be trimmed, as necessary, at a later stage after being glued in place. The braces are made from the same well seasoned Sitka Spruce stock as the sound board. Unlike the braces of modern ouds, the grain of the wood runs horizontal rather than vertical in the brace cross section. This grain direction is typical of 16th C lute practice so may, at one time, have applied to early ouds as well. The horizontal grain provides a better gluing surface to the sound board and may have some acoustical advantage as yet undefined.

The clamping of the braces was done using my 'GoBar' unit. I have had mixed results with this tool in the past. It has worked well with the wider braces found in early guitars, vihuelas etc. but less so with narrow lute braces. The slightest misalignment of the 'gobar' tends to tilt the brace sideways. This is particularly the case with 'gobars' of round cross section that will bend in any direction.
In an effort to correct this problem, a set of 18 'gobars' (enough for clamping 6 braces at one time) were made from relatively straight grained, seasoned Ash wood. The 'gobars' were made flat in section (20 mm wide X 3 mm thick) - cut square at one end and filed to a smooth rounded profile at the other to engage the top of the brace when sprung into place. The intent is to ensure that the 'gobar' will bend only in the direction of the length of a brace and so eliminate any sideways forces that would cause a narrow brace to tilt sideways.

Another potential problem with the 'gobar' system is that the base and top of the frame must be made rigid enough to minimise any deflection due to the cumulative loading presented by the 'gobars when sprung into place. This loading can be substantial, sufficient cause undesirable bending of the frame - not such a problem when clamping two (or three) braces for an early guitar but more so when many more braces are to be clamped at one time.

The braces were glued to the sound board with hot hide glue. Each brace was first heated before application of the glue in order to increase the brief time available before the glue 'gelled' and became unworkable. Working as quickly as possible, the braces were glued and clamped in place. However, the flat section 'gobars' did not prevent three of the braces from tilting slightly sideways under clamping pressure. I was also uncertain about the consistency of the glue which I felt, in retrospect, was insufficiently diluted or 'watered down' for the task.
So, nothing for it but to remove the braces and start again! Not a big problem when using hot hide glue. Using a low cost, thin, stainless steel artists spatula (4 for $10 Can. from Lee Valley), the glue at both sides of each brace is first moistened with a brush dipped in hot water. An electric hair dryer, at maximum heat setting, is then used to heat the glued joint while the spatula is carefully worked into the joint at one end to separate it. At this stage water is then brushed only onto the blade of the spatula which is gently pushed further into the joint (no forcing!), as heat is applied, until the brace completely separates from the sound board.

If relative humidity drops below 55% within the next few days, the braces will be cleaned of glue and used for for a second gluing attempt. If the 'gobar' system does not work satisfactorily then the braces will be clamped individually using a clamping brace.
If prevailing humidity levels exceed 55% then all that can be done at this point in time will be to brace the less critical sound hole rosettes.

Thank goodness I don't have to make instruments for a living!

No - in order to replicate the oud represented by the engraving as closely as possible, the only 'decoration' on the soundboard will be on the bridge and in the sound board edge banding.
Ibn al-Tahhan al-Musiqi in the 14th C. wrote that "the best ouds are uncarved and undecorated and are made from one kind of wood" (meaning the wood of the bowl, I assume). He goes on to write that "If it is desirable that it should be decorated with ebony, this should be as light and thin as possible, and sparsely used" (Dr. G.H. Farmer translation).
So the top of the bridge and the soundboard edge banding on this project will be thin (about 0.5 mm thick) and of contrasting ebony/boxwood tiles (as indicated on the bridge in the engraving).

As wet Spring weather is currently moving through this region (100% relative humidity outside - 60% inside the workshop today) no progress can be made on any assembly work until drier conditions prevail.

The "cut in" rosettes are very fragile at this stage so must be reinforced. This was achieved traditionally in 16th C lutes by gluing thin strips of sound board wood underneath each rosette across the sound board wood grain.
The reinforcement strips have been prepared from scraps of sound board material - thinned to 1.5 mm thickness and 2.5 mm depth - using a purfling thicknesser. This simple tool is made by cutting grooves of varying depths in a block of Maple wood (I used a router to make the grooves). The strips are then all brought to the desired uniform thickness and depth by pulling them through the selected groove while holding a block plane at an angle on the top of the purfling thicknesser (the plane is held stationary throughout this operation).

The image shows the strips, arranged as they will be glued to the back of the rosette. Hot hide glue will be used for this operation so the strips will be held flat and in place - until the glue is fully cured - using 'cauls' made from chipboard to which a soft, flexible synthetic rubber facing has been attached with double sided adhesive tape (The rubber facing is a cheap, 'non slip' underlay used for carpets). The 'cauls' will be held in place using lead weights (wrapped in masking tape to eliminate all possibility of lead poisoning when handled). Better safe than sorry.

The front face of each rosette will then be further consolidated and strengthened by painting the surfaces with thin Shellac that will soak into the wood. Shellac makes quite a good adhesive and will reinforce any otherwise invisible cracks that may exist in the rosette piercing work.

No Alami, for the next oud John will carve the body )
John's patience and discussion of details is amazing. This is a really neat project.

Still too wet and humid to do instrument assembly work (and busy with other non-instrument making priority activities). However, component preparation and fabrication can go ahead - as time permits.

The hot bent rib blanks have now been sorted into their proper sequence for assembly in the bowl and numbered accordingly. The bowl mold has also been numbered to match the rib positions. The rib centre lines were marked on the ribs prior to bending and are now used as a reference for marking the rib profiles. The previously made metal rib template is clamped to each blank - with centre lines aligned - and the rib outline traced in pencil. The rib profile marking is only an approximate guide for rough trimming of the rib blanks that are next to be cut and left slightly oversize.

Conditions have been improving with dry Spring conditions so a 'window of opportunity' for gluing operations presents itself.
Today the temperature is 25 C, sunshine, with a strong breeze blowing. A beautiful late Spring day - tempted to just sit out side and relax but - removed the Winter shutters from my small workshop (temporary conversion from a tinsmith workshop) to allow air to freely circulate. The relative humidity dropped dramatically from 65% to 50% during the morning so decided to go ahead and glue the rosette bracings.

Using the kitchen for gluing operations has been very inconvenient (not to mention justified complaints from my wife who keeps reminding me - in vain - that her kitchen should not be used as a workshop extension or laboratory for luthier experiments). So a low cost, portable, electric two ring cooking range has been purchased to heat the glue pot.
The concern is that in the confined space of the workshop, the steam from the water jacket heating the glue pot will raise the relative humidity of the work space.

My tinsmith operations (involving a lot of soft soldering with toxic fluxes) have necessitated installation of a fume extraction hood - using a high capacity fan to extract the fumes to outside the workshop.
This is a home built unit. The hood is made from 3mm 'Plexiglas" (or Acrylic) plastic sheet formed by softening the plastic with a hot air gun and then bending it to shape. - all glued together with epoxy glue. The compact, high output fan is a plastic 6 inch (150 mm) diameter purchased as surplus stock (ex computer? - from the days when computers occupied whole buildings!). The fan unit has been installed inside a box in the wall of the workshop - venting to the outside via a standard 6 inch stove pipe bend (fitted in a wooden block so that it can be rotated if necessary, dependent upon wind conditions). The connections between the hood and fan are made from standard black plastic sewer pipe fittings - with a piece of flexible, 'laundry drier' tubing connected with hose clips.
This arrangement has worked very well as a soldering station for a decade or so. Testing it with the boiling glue pot confirmed that the steam was effectively removed to outside the building - perfect!

With the workshop glue station up and running, the rosette bracing was installed. The purpose of the bracing is to support the delicate tracery of the rosette against damage and to stiffen the area of the sound hole. The bracing is made from spruce scraps from sound board material - about 1.5 mm thick by about 3 mm deep.
The gluing operation was carried out in a rotating sequence. One brace was coated in glue and held in place for a few seconds - until the glue gelled - and then clamped flat with a weighted caul before moving on to the next rosette.
The completed bracing will be left for a day until the glue has cured hard. The bracing will then be planed to a uniform height (which will, in turn, help test the strength of the glued joints) before moving on to the next step.
If the relative humidity conditions remain favourable, the sound board braces will next be glued in place.

Low humidity again today so went ahead with gluing the soundboard bracing.

The rosette bracing has been planed (in situ) to a uniform depth of about 2.5 mm.

As the gluing of the braces has been delayed for a while - after the unsuccessful attempt previously posted - the brace positions on the sound board were lightly sanded to remove any oxidation that might affect the glue bond. These surfaces and the joint surfaces of the braces were also cleaned with a cloth moistened in propyl alcohol - just for good measure.

The go-bar rig was again used but the Ash go-bars have been modified to reduce the loadings that seemed to be excessive. This was done by reducing the width of each go-bar from 20 mm to 10 mm - except for the top 50 mm which was left at 20 mm width to preserve correct alignment of each bar under load. Go-bar thickness remained at about 4 mm.

It was decided to glue only three braces today (for convenience and to minimise the total load on the go-bar rig) - the remaining three will be glued tomorrow - although with freezing temperatures in the forecast, the workshop heater may need to be turned on. We are not out of Winter yet!

This time around, the go-bar system worked a lot better with no tendency to 'tip over' the narrow, relatively deep braces.
The glue was made a bit thinner - nevertheless, it was still necessary to work quickly to get the braces in position and clamp each one in place with three go-bars before the glue 'gelled'.

Another dry day yesterday so with the workshop relative humidity maintained between 50% and 55% the gluing of the braces was completed. Maintenance of relative humidity at these levels is important because once the bars have been glued in place, the sound board is no longer free to expand and contract across its width with humidity changes.

Today is a different story with rain in the forecast for the next three days.
After drying overnight, the sound board was removed from the go-bar rig and then hung on a peg in another room of the house for safe keeping and for further drying/curing of the glue. At this stage, all of the bars are of equal height of 15 mm. No further work will be done on the sound board until it is fitted to the bowl.
The pressure of the go-bar clamps causes small indentations in the top of the braces. The braces have been made slightly oversize so these marks will be planed out later when the bracing is finally finished (after the sound board has been fitted to the bowl).

With higher humidity levels a start was made today on another component - the bridge.
Three pieces of pear wood were selected from stock and tested for suitability by hand planing the sawn surfaces to reveal any small defects or grain irregularities. One piece was found to be suitable with no grain 'tear out' when planed. The wood is close grained and works nicely under the plane. It has a slight pink colouration with some slight 'fiddle back' grain figure in evidence.

The bridge of the oud in the engraving is very lute like in appearance so the design will follow lute bridge design found on surviving lutes of the 16th/17th C. Comparing the bridge dimensions of my old Arabic oud with those of lutes - the relative dimensions of string spacing and height are almost identical (the oud string height is a bit higher). The bridge on the Arabian oud is wider than found in lute bridges (that are narrower and longer) but the 'foot print' (area covered by each) is practically identical.
The extra length on a lute bridge is provided by the carved, 'decorative' ends of the bridge. Here an early 17th C bridge end design has been modified slightly in order to conform more closely to the oud engraving. This bridge will also have tiled inlays on the front and rear top edges - as depicted in the oud engraving.

Proportionally this design fits quite well with the bridge shown in the engraving.

Small portable dehumidifiers- more than adequate for a small workshop space cost around $200 - they are just little electric powered refrigerators with an air circulating fan and control unit. It would be nice to have but we usually manage to keep cool by opening windows and using fans. The advantage of the cold winters and heated building interiors is that the resultant relative humidity can be good for instrument building. Then we use humidifiers if the air gets too dry (we have one of those)!

Wet, warm and humid today - condensation on cold surfaces - so continued with work on the bridge. The first step is to rough shape the blank and drill the holes with everything 'square'. As the oud is fretted I am following lute practice with string holes lower on the treble side than on the bass (a 1 mm slope). The drilled and slotted bridge ends are first roughly carved with a chisel to remove excess waste material and then brought close to the finished contours (I used a small drum sander in a drill press for convenience).
The top front and rear edges of the bridge will have a tiled inlay - matching the sound board edge inlay (yet to be done). This could be applied as a veneer panel glued to the top of the bridge but I have decided to set the inlay into rebates. The rebates were cut - 1mm deep and 3 mm wide - on a router table (for convenience again). The sound board edge tiles will be 4 mm wide by about 0.6 mm deep.
The tiled inlay must be glued in place before the bridge can be finished so the next stage is to make the inlay.

I shall be using African ebony and Persian boxwood for the tiles - both appropriate materials for an early oud.
I purchased two logs of the boxwood from a timber importer in Liverpool, U.K., during the 1970's (he had a huge pile of the logs in a bin - as well as logs of Indian rosewood, billets of Brazilian rosewood etc. , all sold by weight. Those were the days!) Boxwood is notorious for the length of time it takes to season but this should be OK for this application. This stuff is hard, close grained and is an attractive lemon yellow colour that should contrast well with the ebony. It finishes with a mirror surface straight off the plane - nice wood.

The first step is to prepare a few short sticks of the ebony and boxwood - all 5 mm wide and about 2 mm thick. The edges of each strip were trued - square and straight - on a 'mini' shooting board - ready for gluing.

The Ebony/Boxwood strips have been glued together side by side and both faces of the assembly planed flat and to a uniform thickness of just under 2 mm. During planing, the assembly was held in place on a flat board using double sided adhesive tape.
The inlay strips are made well oversize - about 4 mm wide - to allow for trimming. A line is first scored with a knife as a guide for a fine tooth razor saw. The work must be firmly clamped during sawing to prevent breakage of the strips.
After each strip has been cut the edge of the assembly is made square and straight by 'shooting' with a plane - ready for cutting the next strip. The straight face of the strip will be glued into the rebate of the bridge.

Both Boxwood and ebony are brittle woods so some slight tear out or chipping of the edges is hard to avoid. However, as the inlay strips are made well oversize these flaws will be remove when the bridge is finished to size.
To ensure a close fit in the rebate, the inside corner of each strip has been rounded slightly.

The inlays have now been glued to the bridge and the glue left to fully cure overnight.

Work on the bridge is now complete apart from final smoothing and finishing.

The black and yellow tile inlay made a dramatic contrast against the pale wood of the bridge so the bridge has been given a primer coat of orange shellac as a test to see how it looks (and to seal the wood grain). However, I am not sure that I like the overall orange tint of the bridge so may remove the shellac during final finishing.

The sound hole rosettes have also been coated with orange shellac to penetrate, consolidate and strengthen the thin wood (shellac makes a good glue). The rosettes will be left with the shellac finish - contrasting with the unvarnished sound board wood.

The bridge has been placed (not glued!) in its correct position on the unfinished sound board just to see how it looks. Note that the sound board, at this stage, is about 12 mm oversize all round.

Relative humidity is still on the high side but a start was made on the bowl today - hoping that conditions will improve over the next few days.

The rib blanks have all been pre-bent, numbered in sequence and left in a drying rack - as reported in a previous post. As there is some 'spring back' of the pre-bent ribs, each rib blank must first be slightly adjusted again to the exact profile using the hot bending iron
The profile of rib #1, the central rib, has been marked on the blank - in pencil - using the metal rib template.
One side of the blank is carved close to the line (to about 2 mm) using a sharp knife - working in two directions from the centre point of the rib, following the grain of the wood. Then, using a small block plane, held in one hand, the remaining excess material is cut away down close to the pencil line.
The accuracy of the joint is tested as work proceeds by placing the rib on a surface plate and viewing the rib against the light to view any high or low spots. This is quite a severe test but work must continue until no light can be seen along the length of the rib joint. This method works because the bowl of this oud is semicircular in cross section and the ribs (theoretically) are of equal geometry.
The finished joint surface could be cut using the hand held block plane alone but it is perhaps a bit easier to make the final cuts on either an inverted jointer plane or on a flat sanding board. I prefer to use a sanding board as I can get the joint very close using just the block plane used freehand. It is then just a matter of a few light 'swipes' on the sanding board to remove any slight high spots.
The problems with using a jointer plane is that there is only a relatively small surface area to work with, often the plane surfaces are not flat (poor manufacture) and the rib must be worked in two directions - with the grain of the wood. However, it is all a matter of personal preference and what works best for each luthier.
Once a perfect joint surface had been obtained on one side the process is repeated for the other joint face.

My sanding board is made from two pieces of 20 mm thick MDF (Medium Density Fibre board) glued together to make a flat surface on which sheets of 220 grit Garnet paper (sandpaper) have been glued.

My surface plate is made from a piece of 8 mm thick plate glass mirror. The mirror - which had lost half of its 'silvering' - was given to me by a friend ('waste not want not' - this stuff is heavy and expensive when purchased new). A section that still retained its silvering was cut to size using a standard carbide wheel, hand held, glass cutter. The process is the same as cutting thinner window glass - use a metal square as a guide, kerosene as a lubricant and steady pressure to score the line. It requires quite a force (a heavy fist!) to break the glass along the scored line. As the resultant cut edge was not absolutely square and has sharp edges (I am not a professional glass cutter with diamond edge grinding equipment), the glass was placed on a piece of
20 mm MDF with some pine strips (slightly thicker than the glass) to hold it in place and to avoid any risk of cuts to the hand when handling.
The mirror's silver backing is coated with a matt black paint so this forms the accurate, flat working surface of the surface plate with the mirror placed face downwards.

The completed central rib is placed on the mold to check the accuracy of its geometry.
The front end of the rib is held in place with a pin (though a drilled hole to avoid splitting the rib). This hole will eventually be covered by the bowl end plate.
At the centre section, the rib is held in place using a rubber strap (a strip cut from an old tire inner tube - waste not want not!).
For convenience of working, the mold has been fixed to an adjustable base.

If conditions are favourable tomorrow the central rib will be glued in place so a fresh batch of hide glue has been prepared in anticipation.

The central rib was glued today. With a week of sunshine in the forecast some progress with the bowl may be possible.

The rib has been fixed in the correct position on the mold using small blocks of pine pinned to the mold bulkheads.
At the neck block the rib has been clamped in place with plastic push pins - after application of hot hide glue. Small card squares underneath the pins are to prevent damage to the edges of the rib.
At the tail end, the rib has been pinned to the end block of the mold. The end block has been covered with 'oven proof' paper to prevent the ribs being glued to the mold.

Apart from the rubber band which retains the rib to the mold at the centre - no force is necessary to make the rib conform to the mold profile.

Perfect weather conditions - a shame to spend time indoors - but went ahead with rib #2.

The technique for gluing the ribs is the centuries old method described by Arnault de Zwolle in the mid 15th C - still used by some oud makers like Dincer Dalkilic working in the Turkish tradition.
Using hot hide glue, the rib joints are held in place with glued strips of paper scorched with a hot iron (to cause the glue on the paper to set quickly and also to remelt the glue in the joint to obtain a close fit of the ribs). I have never used this method before so this is a 'hands on' learning process.

The second rib has been prepared so that it will fit against rib #1 and the mold bulkheads without being forced into place.

The glued paper strips are freshly made from old, brown paper envelopes - coated with hot hide glue and allowed to dry partially before being cut into pieces measuring about 25 mm X 20 mm.

The joint surface of rib #2 is first coated with hot hide glue, then - starting at the neck block end - the rib is glued to the neck block. The glue is remelted with a hot iron and clamped in place with two pins. By this time, of course, the glue applied to the joint surface has 'gelled'. Any surplus glue on the face of the rib can then be rubbed off to avoid making too much of a mess. The hot iron is then worked over the rib joint - remelting the glue and closing the joint surfaces. While held in this position, with light side pressure, pieces of glued paper are applied over the joint and quickly sealed in place with the hot iron. This process is continued until the whole rib joint has been glued.
The paper strips will remain in place until the bowl is complete, released from the mold, and the joints reinforced from the inside with glued paper.

To prevent the ribs from being glued to the mold, oven heat resistant paper has been placed under each of the bulkheads. However, any ordinary scrap paper would serve just as well.

The thermostat of the hot iron was set at half of maximum temperature.
The iron need not be a fancy electrical device - a home made iron - a piece of copper with a handle - heated over an electric stove ring or propane burner would serve just as well - but less convenient.

So far so good!

Steady progress with rib #5 now being fitted. With the current fine weather there is only time to devote to fitting about one rib a day as other work takes priority.
An advantage of the glued paper/hide glue method of assembling a bowl is that the paper strips, when scorched in place, immediately hold everything securely together allowing work to proceed on the next rib. No need then to first wait for glue to dry - as would be the case if modern synthetic adhesives are used. It would be quite possible for an experienced luthier - working to a tight schedule - to assemble a complete bowl, using this method, in a day or two.

Still gaining experience with this method.
Several glued paper strips were left over from fitting ribs #2 and #3. As the glue on the paper dries and hardens it shrinks causing each paper piece to curl up. Rather than throw these away, a test was tried on rib #4. By quickly dipping each of the strips in water prior to use the paper strips flattened out as the paper absorbed the moisture. However, the moistened paper also loses its strength so it was found to be more difficult to 'iron' the strips over the rib joint - the paper tending to stick to the iron and tear before the glue dried out.
So, the best approach is to freshly prepare the glued paper at the start of each rib gluing operation - this way the paper strips remain flat and retain their full strength when being ironed into place.

Thanks Leon - I have some further slight modifications to make to the bridge - apart from removal of the orange shellac - that will improve functionality of mounting the strings and will remove excess material to further reduce the weight of the bridge (slightly). But more on that later.
Another nice day so rib# 5 glued in place and rib #6 being fitted.

In a previous thread last year ("Early Oud Construction Methods?"), I ran some experiments to test if the glued paper procedure resulted in the rib joints being pulled together as the paper shrank (across the 'grain' of the paper) when scorched in place with a hot iron. The paper strips in these trials were first soaked in water - after which hot hide glue was applied before each strip was place over the rib joint and scorched to instantly harden the glue. It was concluded that any shrinkage of the paper during scorching had already taken place by the time the glue had hardened so that no significant tension was applied to the joint by shrinkage of the paper.

Another trial was undertaken today to test if glue applied directly to paper - without the paper being pre-soaked - resulted in shrinkage of the scorched paper strips across the rib joint - to shrinkage of the hide glue as it dries.
The test simple test rig was made from two strips of spruce - about 10 cm long - freely hinged at one end and with a gap of 0.025 inch (0.63 mm) - accurately set with an engineers 'feeler' gauge at the other end.
Sample A - the paper was scorched in place with grain direction of the paper parallel to the joint (i.e. the 'curl' direction of the strip is across the joint). After scorching the joint was found to have closed by about 0.005 inch.
Sample B - with the curl of the paper along the joint - there was no measurable closure of the joint after scorching.

Conclusion to follow.

Humidity still on the high side but decided to glue rib #7 to avoid losing momentum. Only four more ribs to go.

When gluing the ribs a start is made at the neck block. This end of the rib is clamped in place with two push pins before continuing - bit by bit - along the remainder of the joint with the hot iron and glued pieces of paper.
The final operation is to remove the two pins at the neck block. The area of the rib over the neck block is then ironed with the hot iron to re-melt the glue and the pins replaced to hold the rib in position.

To avoid damage to the edge of the rib, small squares of card are placed under the hard plastic head of each pin. It is possible also for the metal pin itself to dent the side of the joint. In this case any slight dents can be easily removed later - after the glue has dried - by brushing a small drop of water over the area of the dent and applying the tip of the hot iron. This causes the wood to swell and removes the dent - as can be seen in the lower image. It is possible that a pin dent might also be removed when ironing the joint during gluing - due to the moisture in the glue and the applied heat - but I haven't tried that.

Humidity levels continue to fall (but it will not last as more rain is in the forecast) so decided to move ahead with fitting and gluing ribs # 8 and #9.

I miscalculated the number of glued paper strips required for two ribs so supplemented the freshly prepared strips with some dried and curled strips left over from yesterday. Instead of dipping the dried strips in water, each glued area of each strip was first wiped on a moistened rag. This was sufficient to 'rejuvenate' the glue without weakening the paper when ironed in place. Strips rejuvenated in this way have lost most of their shrinkage advantage but, nevertheless, help to hold everything together.

By the time rib #9 had been glued in place, the paper strips, freshly prepared at the start of the operation were beginning to dry out and curl. So - at the speed that I work - it is best to work one rib at a time and prepare sufficient glued strips for each rib.

The outside ribs # 10 and #11 have been made, by design, a bit wider than the other ribs. This is an allowance to compensate for any accumulated errors that may occur in the fitting of the ribs. However, as rib geometry has been kept under control during the fitting process, the outside ribs will remain slightly wider than the rest.

Another beautiful, sunny late Spring day, with humidity levels dropping to acceptable levels, so ribs # 10 and #11 have been fitted and glued. Mission accomplished!
The ribs are a close fit to the mold profile.

The outside edges of the ribs were marked with compasses using the base of the mold as a reference - allowing about
5 mm extra width for any trimming later. This marking will be used as an approximate reference line only.

After allowing the glue to dry overnight, the bowl will be released from the mold tomorrow - hopefully without any problem!

The next step will be to, first, glue in the reinforcing end plate and then the paper or fabric strips covering the interior rib joints.

The bowl was released from the mold without problem.
The measured outside dimensions are width 34 cm and length (neck joint to bottom of bowl) 45 cm - exactly to design.

These dimensions may change slightly after the end plate and interior rib joint reinforcement are added - due to glue shrinkage.

The end plate reinforcement has been modified from the end plate in an old Arabic oud - due to the spherical geometry of the bowl (the Arabic oud is more flattened in profile). This design of end plate conforms to 16th/ 17th C lute practice which, in turn, may have come from early oud design.

The end plate has been made from Ash (same material as the bowl) - hot bent to the bowl contour. It has been made 1.5 mm thick at the upper edge tapering to a 'feather edge' of about
0.5 mm. This provides flexibility at the edges to closely fit the bowl contours.
After several practice 'dry' runs, the end plate was glued in place with hide glue. Speed is of the essence here - as the glue sets in less than a minute - so the spring clamps were ideal for the job.

While the glue on the end plate is curing, the rib joints were reinforced with fabric strips. The strips have been cut from Chinese raw silk fabric (recycled garment cloth) - 15 mm wide - glued in place with hide glue. The procedure is to first coat each joint with glue, stick the fabric in place and then brush on more glue to fully saturate the fabric with glue.
As there was only enough glue to complete five strips, a fresh batch of glue has been prepared to finish the job tomorrow.

With the remaining silk, rib joint, liner strips glued in place this morning the time has come to remove all of the glued paper strips from the exterior of the bowl.

Not sure of the best way to go about this so consulted 'Musick's Monument' by Thomas Mace, London, 1676 (see image). Mace describes how to remove and replace the sound board of a lute as a regular maintenance activity. The sound board is clamped in place, on the bowl, with glued paper strips scorched with a hot iron. To remove the strips, Mace recommends moistening the strips with a wet cloth (to soften the paper and glue) after which the paper is scraped away using fingernails (to avoid damage to the varnished surfaces of the bowl).
Testing a small section of the bowl according to Mace's instruction - the glued paper strips eventually softened sufficiently to be scraped off but the process was very slow and messy. There must be a better way!

Carefully slicing off the dry paper strips with a paring chisel was also very slow and there was the risk of cutting into and damaging the surface of the ribs. So, no improvement.

The most effective method found for removal of the glued strips was to place a moistened cloth over the strips and apply the hot iron for a couple of seconds. This was sufficient to immediately liquefy the glue in the strips so that they could be easily removed with a plastic (or wooden) scraper - working with two or three strips at a time. It is necessary to move quickly in removing the strips as the glue will re-harden within a few seconds.

The remaining glue residues on the surface of each rib - as well as the wood grain 'raised' by the moisture - will be removed and finished with a curved scraper blade after everything has been allowed to dry overnight.

Now that the external paper strips have been removed and everything has dried the time has come to clean up the bowl exterior to remove all glue deposits, pencil reference marks and other minor discrepancies.

This is work is all done with a scraper blade - which requires use of two hands - so the bowl has been placed back on the mold (now imperfectly fitting due to the addition of the end plate) and held in place with the rubber strap. The scraper blade used is from the Lee Valley Tools set of 'super hard' curved scraper blades (the' barrel' shaped blade from cat# 05K20.20 or 05K20.10 that happens to nicely fit the the natural fluted curve of the ribs that I wish to conserve for this instrument).
This work requires some care and attention so will take a few hours to complete so - with other priorities - will likely be completed in the next day or two. All the time in the world - but not a moment to lose!!

As a matter of interest, the bowl, soundboard and bridge, neck blank (and a six course peg-box assembly) have been individually weighed from which it can be estimated that the weight of the completed oud will be about 1.5 pounds or 0.68 kg. I am aiming for an even lighter weight than this - but time will tell.

The final component to be fitted to the bowl is the clasp (not sure what this is called in oud terminology).

On a modern oud this is often just a semi-circular disc of wood - about 50mm or so in diameter that is primarily cosmetic in function - to cover the imperfectly fitting tips of the ribs and any nail holes that were used to hold the ribs in position over the tail block during construction of the bowl.

On surviving lutes of the 16th and 17th C the practice (which may also reflect earlier oud construction) is to reinforce the bottom perimeter of the bowl with a clasp - glued to the outside of the bowl - and usually extending as far as the widest point of the bowl. The clasp not only covered up the tips of the ribs and any nail holes but provided additional structural stiffness to the bottom of the bowl and to the sound board in the area below the bridge.
The lute clasp was also decorative and the terminal points were either left plain - cut simply at an angle - or provided with a more fancy scalloped motif. The attached image shows some examples of clasp points sketched, to scale, from surviving lutes of the 16th C. This tradition and design detail continued to be a feature of 17th C lutes.

After gluing the silk liner strips and the interior end plate to the bowl, the width of the bowl - as predicted - has reduced slightly by 5 mm to 33.5 cm due to glue shrinkage etc.
By using an end clasp, it should be possible to apply a slight corrective pressure to the bottom of the bowl sufficient to restore the original design dimension of the bowl to the maximum design width of 34 cm. So a lute like (or early oud like?) clasp it is!

The template for the clasp is first made from card to test the best fit to the bowl.
The second image shows a first attempt. The design of this clasp point reflects the motif of the ends of the oud bridge which I think is the way I want to go. Card is cheap so will make more trial attempts in order to optimise the design before cutting wood. I have a nice, but modestly, figured piece of Ash wood left over from the rib stock that I think would do for the clasp

The clasp will not extend to the widest point of the bowl - as is usual practice on surviving lutes - but will only cover the same perimeter distance as the internal end plate. The bottom of the bowl as far as the first brace is, therefore, 'sandwiched, between the end plate and clasp to give an overall wall thickness of about 5 mm. The sound board banding will be
4 mm wide so - unlike the remainder of the sound board - will not be effective in this area as an edge weakening device designed to modify sound board vibration and response.
This method of providing stiffness to the sound board below the bridge is similar to that used on my old Egyptian oud which has a thin interior end plate (but not a clasp). However, instead of using a clasp, additional stiffness is provided in this area by omitting the edge banding and also by covering the sound board edge with a leather strip glued in place.

Compare this to the method used in the 15th C Arnault de Zwolle lute where the required stiffness is provided simply by employing a very thick tail block. This method could have been used on this project but I wanted to test how a lighter and more sophisticated solution might work.

Modern ouds, of course, generally have a much thinner tail block (due to a more flattened bottom profile of the bowl) together with a brace below the bridge - a design solution falling between that of the Arnault lute and the oud of this project.

The final shape of the clasp has been established, by trial and error, using card templates.
The Ash wood blank has been planed to a thickness of under 1.5 mm and the profile marked in pencil by tracing around the card template. The decorative clasp ends are fragile and will easily break (due to the wood cross grain) so the 'scallop' motif has been cut first using a sharp, single spur, cutter in a drill press - the work being clamped securely for this operation to avoid risk of breakage.
The profile of the clasp was then carved close to the marked line with a sharp knife and finished to size with files.
The top of the clasp has been left about 10 mm oversize. This will allow some latitude for obtaining the best fit on the spherical surface of the end of the bowl.
The clasp will be hot bent and fitted with a slight amount of 'spring' to bring the width of the bowl back about 5mm to the required 34 cm design width.

Thanks Peyman. What a charming miniature painting - is it Persian? An interesting combination of instruments - harp, pipes, tambourines and oud. I love the depiction of the 'cross eyed' piper behind the oud player - it must have been quite an effort to blow those things!
So again we have an example of the sickle shaped peg box, and - unusually(?) - a clear representation of contrasting tiled banding around the edges of the sound board, finger board and pegbox. Are those frets on the fingerboard- or just decorative panels? Rather late for frets on an oud? Also note again the mysterious two tone colour of the sound board - what can that mean?

Humidity in the workshop (65% RH) is too high today for assembly work so spent time hot bending and fitting the clasp to the bowl. The clasp is flat but the bowl is spherical so it was necessary to glue a thin wedge of Ash wood to the central part of the clasp to obtain a close fit all around. In order to apply slight pressure and bring the bowl width back - 3mm - to the design of 34 cm, a flexible 'strong back' of spruce is to be used under the central clamp - verified by a 'dry' run. When the clasp is glued in place, the bowl width should be maintained close to design width.

Continuing, uncomfortable, high humidity levels so spent a little time modifying the bridge.
The shellac coating was removed and the bridge ends re-profiled and reshaped to more accurately reflect the bridge design depicted in the engraving. These modifications resulted in a weight reduction of the bridge from 7 grams to 6 grams without any significant reduction in the bridge 'foot print' (area glued to the sound board).

At around 3 pm, the 'winds of change' in the weather conditions could be felt blowing through - so hopefully lower humidity conditions will prevail over the next few days.

Current hot, but rain free, weather conditions over the past couple of days mean that Relative Humidity is around 64% in the morning dropping to around 50% by mid afternoon as the moisture in the atmosphere is 'burned off'. So the clasp was glued yesterday afternoon - to complete the bowl structure - at an RH in the workshop of 53%.
Digital 'Weather Stations' - measuring inside and outside temperatures, as well as indoor RH, were on sale at low price at the local hardware store, so picked one up to replace the less accurate, cheap analogue unit that I have been using so far. (Subject of another thread yet to be posted - for information).
The edge of the bowl must now be trimmed to the level of the neck block which means that about 5 mm must be removed by planing. The correct level is determined by placing the bowl on a flat MDF working surface and tracing around the edge of the bowl using a pencil taped to a piece of wood of appropriate thickness. Simple, 'make shift' tools that do the job.
The slight tension applied to the clasp before being glued in place has brought the width of the bowl to within 2 mm of the design 34 cm - close enough.

With the bowl structure complete, the rib ends overlapping the neck joint face have been trimmed and finished flush with the joint face.
The trimming is best done with a sharp knife making shallow carving cuts across the wood grain. To finish the joint face, the neck block is firmly supported on a small board - clamped to the workbench - and filed flat using a fine, double cut, machinists file. A file is a cutting tool that can give more accurate results than a sandpaper block which, nevertheless, might also be used as an alternative. Beware of 'not so inexpensive' files that may be twisted and warped and incapable of producing accurate results. Pay a little bit more for good files from reputable manufacturers like Nicholson.
The width of the finished joint at this stage is 5.1 mm (5 mm is the minimum width by design).

The neck blank dovetail (prepared during the winter period) has now been precisely shaped and fitted to the neck block.
I have not used this technique before but the great advantage is that it accurately locates the neck in relation to the bowl allowing any necessary alignment adjustments to be made prior to final gluing of neck to bowl.
The 'disadvantage' is that any slight discrepancy in the dovetail or neck block alignment is greatly magnified over the length of the bowl or neck. For this reason, I have left extra material in the neck blank and bowl for final trimming and adjustment.

The neck dovetail has been reduced to the required, close, dimensions using a paring chisel - checking accuracy, as work proceeds, with a good quality steel ruler. The final, precise, fit is made using a file - carefully removing material, little by little. The whole process took about an hour to obtain a precise fit.

Although the dovetail fits accurately, the centre line of the neck now requires about a 2.5 mm adjustment towards the bass side. No problem - as there is ample material left in the neck blank to accomodate this adjustment.

Shrinkage of the hide glue in the inter-rib liner strips has also caused slight backward rotation of the neck block so that the neck block upper face (and the neck) tilt back from horizontal. This may be corrected (without applying longitudinal stress to the completed bowl) by gluing a small wedge of spruce to the neck block and neck or by making the adjustment in the bowl (where the extra 5 mm edge allowance may be sufficient). I prefer the latter course of correction.

To be on the safe side, additional material has been glued to the upper surface of the neck block (about 4mm at the thickest point tapering to zero at the dovetail). This gives maximum latitude for adjustment - although most of this additional material will be planed away in the final fitting and leveling process. Additional work - punishment, no doubt, for not being diligent in checking the flatness of the mold base plate in the first place!!
The dovetail and neck joint surfaces - after careful filing - are now a perfect fit.
The neck blank will be left, unshaped, with square sides, until the bowl leveling process is completed. This makes it easier to clamp the neck block in a vice when planing the upper surface.

Too humid now to undertake assembly work so this may be a good time to turn attention to the design of the pegbox.

The peg box, depicted in the engraving, is something of a mystery (compared to modern day oud construction). Obviously, it is not an exact, perspective, representation of what the artist saw but is subject to artistic convention. So what does the artist tell us?
1- There are 10 pegs, consistent with a 5 course oud - made from a dark wood or light wood stained black. The pegs have relatively long shanks and circular heads - possibly with small, decorative finials.
2 - The peg box is uniformly curved along its length, or is 'sickle shaped' (side view).
3 - The peg box is tapered from the nut to the 'tail' end (side view).
4 - The peg box is set at a more acute angle than 90 degrees to the neck (side view).
5 - There is a mysterious termination at the tail end of the pegbox (side view?). This detail appears in some early Persian miniature paintings. Was it purely decorative or did it have a practical function - perhaps a place to tie a shoulder strap to support the oud when being played 'walk about'?

Modern oud peg boxes are invariably 'S' shaped. However, early ouds not only had curved, 'sickle' shaped peg boxes but some also had straight side, 'lute like' peg boxes. Also, some. according to the iconography, were mainly straight sided except for the tail end - which was sharply curved and terminated in one of those mysterious finials.

Investigating further, the attached image shows the geometry of three test examples of five course peg boxes drawn to scale.

Image A represents a lute like straight sided version (or rather early oud like?). This peg box is set at 90 degrees to the neck (or close to that angle). Note that the strings are contained within the framework of the peg box but that the string clearance above adjacent pegs diminishes toward the tail end.
As this successful design is invariably found in surviving lutes, why were many old oud peg boxes curved like a sickle? Was it simply cosmetic (purely decorative) or was there a practical, functional purpose?

Image B shows a peg box - drawn to the same scale - where the string clearance above each adjacent peg is equal throughout. This results in a curved profile but with peg spacing increasing slightly towards the tail end.

Image C shows a peg box with the same constant string to peg clearance but, with the peg spacing maintained at equal distance throughout, is more curved along its length.

In examples B and C, the strings are not contained within the peg box frame work but fall outside - harp like. Therefore, neither of these alternatives would seem to be appropriate.

The plan is to design the peg box based primarily on a logical and practical geometrical principal rather than cosmetic appearance.
A development of layout C is to keep the string clearance above each peg constant with equal peg spacing. This geometry also produces a curved peg box profile.
The image shows a preliminary peg box layout, drawn to scale, with a peg spacing of 13 mm and constant string clearance of 2 mm - the strings being contained within the structure of the peg box.
The final design may end up being a bit thicker in profile at the nut and somewhat thinner at the tail end - but generally, to my eye, this seems to be the way to go.

The proposed decorative end of the peg box is in the shape of a 'tear drop' (the geometry being exactly that of the oud sound board). This is shown here with a flat upper surface but which could also be curved in profile - as represented in some of the early Persian miniature paintings previously posted. This detail will be made from ebony with - perhaps - a matching central boxwood inlay within the 'tear drop'.
The 'tear drop' could be set at a lower angle relative to the peg box profile but would then be more vulnerable to damage.

The overall length of the peg box turns out to be 170 mm which is the depth of the oud bowl. This is a geometrical relationship found in early lutes.