Old Oud - New Project

John,
it's great that you're going to make this long anticipated oud.
We are going to see and hear a replica of an early oud, to my knowledge it's a first.

looks like 2009 starts to look an interesting and promising year...

I am here!
getting ready to follow this develop.
let me know if you want me to come over... I will make the pot of tea.
as long as I get to watch!!

Thanks for your interest and comments everyone. Winter arrived about a month early in this part of the world with significant snowfall and cold, below 'normal' conditions starting in mid November and now with temperatures below minus 20 Celsius for the last couple of days. If this is a sign of the winter to come then there will not be much woodworking progress until Spring - although temporary conversion of my small, heated metal working workshop attached to the kitchen is a possibility.

The string spacing - regardless of use of plectrum or finger style - will depend upon physical considerations of the player so there is no hard and fast ruling. For example, my lute, played finger style, is currently set up with an overall string spacing, over 5 double courses, of 32 mm at the nut and 65 mm at the bridge. A typical equivalent spacing, over 5 double courses, at the nut for a modern oud would be around 30 mm at the nut and 65 mm to 69 mm at the bridge. However, the string spacing on surviving lutes can be narrower than either of these - often necessary to be able to accommodate some of the more extreme left hand stretches/finger positions called for in lute music of the 16th/17th C.

One of the first considerations will be how to scale the dimensions of the original woodcut - complicated somewhat by the low resolution of the image. Although the geometric profile, bridge position etc. may be quite accurate (see 'Analysis of an Early Oud Woodcut), there has always been some doubt about the neck and fret depiction. It is thought that these are only intended to represent the overall appearance of the instrument i.e. with seven frets set part way along the fingerboard, 5 double courses and sickle shaped pegbox with an odd looking finial. The width of the neck looks more lute like (judged from a modern perspective) rather than oud like.
If the oud engraving is scaled using an overall string spacing at the nut of 30 mm and 65 mm at the bridge, this gives a string length of about 35 cm if the replica oud is to look like the engraving. That would make it a small oud (although a bit larger than surviving mandolinos (descant lutes) that were played either with a plectrum or finger style). Therefore, to accomodate a longer string length, the relative geometry of the neck will need to change - dependant upon the string length chosen. The shorter the string length, the closer the finished instrument will be in appearance to the engraving. Here the chosen string length will be between 48 cm and
57 cm - so it will be necessary try out various string lengths to see how they look.

Incidentally, the modern day 4 course, fretted Tunisian oud with its relatively wide string spacing and wide bridge might be a survivor of the style of oud that is under consideration. So, any detailed measurements taken from these ouds would be interesting for comparison.

Here is a sketch showing the geometrical and barring layout (previously posted on Analysis of an Early Oud Woodcut) with a possible alternative neck joint location as defined by the Arnault de Zwolle lute geometry. Instead of radius R5 defining the neck joint location, R3a gives a slightly higher location at H which results in a narrower width of the fingerboard at the neck joint (and shallower depth of the neck at that location). If the assumption is then made that the neck length is 1/3 of the string length (as it generally is on modern ouds - a tradition handed down?) then this coincides nicely with both the centre of the imaginary large sound hole and point H. So simple that it must be a possibility.
The revised neck geometry - with string positions for string lengths of 48, 54 and 57 cm, and an overall string spacing of 32 mm at the nut and 65 mm at the bridge - is shown against the original neck profile. It can be seen that the original neck profile is too wide to be practical with those string lengths. My draughtsmanship of the neck leaves a bit to be desired but it should be clear enough. Note that the fingerboard width is scaled to 4 cm at the nut. At the neck joint - assuming a semicircular neck section - the depth will be about 2.7 cm which is acceptable.
With this string geometry it can be seen that the original bridge is proportionally a bit too long so must be reduced in length accordingly in order to look right.

Overall, it can be appreciated why the bowl geometry of ouds developed into longer, more almond shaped profiles still seen in some of the older surviving ouds of the late 19th/ early
20th C. - but building one of those will be another project!

Now to choose a string length. A 48 cm string length will tune to a maximum pitch of about a' (at A440 pitch standard) using silk or gut strings. 54 cm gives g' and 57 cm f#. Higher pitches can be achieved using modern nylon strings and metal overspun basses. However, as the A440 pitch standard did not come into force until 1939 anything goes for an oud of the 13th/14th C! A treble oud in a' would be interesting.
I am open to suggestions.

Physical constraints will dictate string length. Scaled at 54 cm string length gives a soundboard maximum width of 33 cm which seems about right - so 54 cm it is.

The oud profile and relative length/width of neck looked similar to the well known painting of a 5 course lute by Lorenzo Costa (1459 - 1535) so, for interest, the layout sketch was rotated in perspective view for comparison. I didn't spent much time on this so didn't get perfectly matching perspective angles (and the perspective of the painting is not quite correct) but it can be seen that the oud and the lute compare quite closely if the 54 cm string spacing at the nut applies and the length of bridge is shortened in proportion.

The painting is quite interesting as it shows the right hand position that resulted during the transition from plectrum (risha) style to the new finger style of the late 15th C. Here the first finger and thumb take the place of the plectrum - the thumb striking downwards and the first finger upwards. Also note the thumb of the left hand projecting along the side of the fingerboard.
The four wooden frets on the soundboard are unusual.

The attached image shows a drawing, to scale, of the oud with a string length of 54 cm. The width of fingerboard at the nut is 40 mm which will allow for a slightly wider string spacing than 32 mm (if required) due to the relatively thicker bass strings of silk or gut. The neck joint is at the revised H position so that the fingerboard length is 1/3 of the string length. The neck has a semicircular section with maximum depth at the neck joint of 27 mm. The neck joint is vertical. The bridge has been shortened to maintain the proportions shown in the original engraving.
The bowl is semicircular in section, 16.5 cm maximum depth, and will have 11 ribs with no fillet strips between the ribs.

A decision has to be made about the number of soundholes - either two, as shown on the engraving, or three - including the 'imaginary' sound hole assumed in the layout drawing. Either arrangement will likely work. Has anyone ever come across a picture of an early oud with only two sound holes?

Thanks Samir.
To 'complete the picture' and for general interest, the attached image is a drawing of an oud scaled exactly to the proportions of the engraving, assuming a string spacing at the nut of 32 mm. The string length is 325 mm so it represents a little descant oud. Is there any evidence of small ouds of this size ever being made?
Such an oud would still be a practical proposition as it comes close in size to the tiny mandolino lutes popular during the 17th/18th C.
It would be fun to build an oud of this dimension - just to see how it turns out so will plan also to make one to this drawing. If nothing else, it will be a conversation piece.

Just done a quick sum. A string on my 60.5cm lute playing A (440) would play very close to B for the same tension and 54cm length. Assuming it's the same for the other strings, (and I haven't made a fool of myself) that makes an instrument a whole tone above the normal Arabic oud. Could be good. Putting a capo on my instrument I find it works as well as ever.

Great project

More simple sums: a 54cm lute/oud using the same strings would tune a whole tone higher than a 60cm instrument with tensions no more than one percent different for any string.

string length
60cm 54cm
frequency nearest note difference
C 523.3 581.4 d 587.3 1.01%
G 392.0 435.6 a 440.0 1.01%
D 293.7 326.3 e 329.6 1.01%
A 220.0 244.4 b 246.9 1.01%
E 164.8 183.1 f# 185.0 1.02%
B 123.5 137.2 c# 138.6 1.01%


A 45cm string length would be fun to try, given infinite time and skill. The highest string would be f, then c, g, d etc .., like a sopranino recorder to a descant.

40cm would give g,d,a,e...

We'll all be playing mandolins.)
(table won't align, sorry)

So far the relative dimensions of the oud have been scaled up from the engraving overlay so there may be some geometrical errors. The oud has, therefore, been drawn to full scale for a string length of 56 cm. A revised sketch of the geometry is attached for reference.
It is clear from this work that the previously proposed location of the neck joint B given by an arc R5 (as on the Arnault de Zwolle lute) does not quite agree with a geometry of the fingerboard length being a third of the string length. By further examination of the full scale layout, two more geometric relationships have been discovered.

The first is that if the distance from the centre line of the small soundholes to the centre of the brace above (i.e. distance FG - which is the same as GH) is a value 'X' then the distance from the bottom of the bowl L to the first brace (shown dotted) is 'X' and the distance to the front edge of the bridge JL is twice 'X'.
The bridge front edge dimension JL is also a sixth of the length of the bowl BL - which corresponds with the Arnault lute geometry.

Secondly, measuring from the inside of the bottom of the bowl, the distance to the centreline of the 'imaginary' large sound hole EK is 7 times 'X', KX is 9 times 'X', the distance to the inside edge of the neck block KC is 11 times 'X' (which coincides with position C given by arc R3) and the distance to the revised neck joint position KB is
12 times 'X'.

The locations of bridge, sound hole centres, braces and neck block remain in close agreement with the Mersenne lute proportions - to within a couple of millimeters throughout.

The slightly higher neck joint position results in a reduction in the width of the neck joint to a more acceptable 53 mm - the depth of the neck then being 26.5 mm for a semicircular neck section.
After making all of the necessary small adjustments to the full scale drawing, the overall string length now comes to 56.5 cm which is close enough to what is required.

Work can now commence on making the mold - but not for the next few days as temperatures will be dropping to minus 36 C overnight - too cold to operate the bandsaw in my main workshop.

Temperatures warmed up today (currently minus 1 C) following a major snowstorm so took the opportunity to dig my way into my main workshop and use the bandsaw to cut the base for the mold.
The mold profile was laid out - including centrelines and bulkhead positions - on a piece of 15 mm thick plywood from my scrap box. The profile was made 2mm undersize all round to allow for rib thickness. String length is to be a nominal 56 cm.

Cheaper grades of plywood may contain voids as seen here - exposed when the profile was cut. No problem - the void has been filled with a scrap piece of wood and will be trimmed to size after the glue has set.

To facilitate accurate layout of the mold bulkheads, a pattern has been made on a piece of thin sheet metal (standard tinsmith pattern making practice). The 11 rib positions (defined as radial lines) have been determined by trial and error using dividers. The outside ribs have been made a few mm wider than the remainder.
The radial reference points have been pierced through the pattern with the point of a scriber. In use, the pattern is placed on the bulkhead blank and the reference points transferred to the blank - through the holes in the pattern - using the point of the scriber. After removal of the pattern, the transferred reference points, marked in the surface of the blanks, are then connected using a straightedge and marking knife.

The design of the mold will be the same as that given by Henri Arnault de Zwolle in the mid 15th C that is with 5 bulkheads except that an additional bulkhead is to be added at approximately the centre of the sound hole position - for a little more support of the ribs in that location. This is not essential - the most important area of support being at the bottom of the bowl where the rib curvature is greatest. For this reason Arnault de Zwolle spaces the bulkheads in this area closer together.

The central bulkhead of the mold has been laid out on a piece of 15 mm thick pine (pine is easier to work with than plywood) using the metal template and trammels (beam compass) to scribe the circumference. For an 11 ribbed bowl, the maximum rib width will be about 50 mm (2 inches) with the outside ribs being a bit wider at around 55 mm. With these relatively wide ribs that will be less than 2 mm thick (dependent upon the specific gravity of bowl material chosen), the outside profile of each bulkhead will be flattened at the rib positions (according to Arnault de Zwolle's instructions) - the bowl being constructed like an early lute rather than having the smooth rounded profile of a modern oud. (The early Arabic writers indicate that the ribs of an oud should be made thinner than the sound board).
It is also likely that ribs of this width and thickness will become slightly fluted after bending but no compensation for that is currently to be made in mold design unless the fluting turns out to be more significant than anticipated.

To my eye the kanz al tuhaf oud (if that is where the image originates) has a sharp transition between neck and bowl (i.e. a neck joint) so is not comparable to the Sassanid instruments represented in the metalwork artifacts. These have a smooth transition between neck and bowl implying that they are carved from a solid piece of wood (whereas a neck joint implies that the bowl is built up from staves). The crude representations illustrate the difficulty of being able to obtain much useful information from very early sources. To attempt to reconstruct an instrument based on these images would be futile - pure guesswork. Nevertheless, some early images may have some potentially useful information - like the sickle shaped pegbox of the kanz al tuhaf drawing already mentioned.
The attached images show oud(?) players depicted on Sassanid era (6th and 7th C) metal plates (interestingly, both performing under grape vines). Both instruments have the sloping shoulders typical of instruments carved from the solid (assuming that the bowls are round and not flat backed). The image of a European 15th C Gittern (i.e. not a lute) illustrates quite well the carved body.
However, discussion of musical instruments of the 6th C - although of general interest - is off topic. So to return to the 14th C and the oud that is the subject of this thread.

Farmer refers to, and translated from, an original copy of the kanz al tuhaf in the British Library shelfmark Or 2361. He gives the dates of various copies of this manuscript as 1346, 1355 and 1362. He also gives the name of the author as Muhammad al Alumi that is presumably marked on his copy of the manuscript? He states that three 14th C writers mention the use of frets (dasatin) on ouds - Quth al-Din al-Shirazi, Ibn Ghaibi and the author of kanz al tuhaf. He also states that he could find no evidence of frets depicted on hundreds of pictures of the oud dating from the 13th C to the 20th C.
Peyman, can you give more details of your reference source of the kanz al tuhaf? Is it a modern facsimile of the entire manuscript or an incomplete modern edition only providing extracts from the manuscript with pictures from other non specified sources?
I have tried to trace the manuscript Or 2361 on-line at the British Library but it is not listed among the Reference section so may only be accessible to researchers with privileged access visiting the Library reading rooms in London.

The engraving of the oud that is the topic of this thread is supposed to be printed in one of the kitab al-adwar manuscripts written in 1333-34 although this has yet to be verified. The British Library lists two modern publications of the manuscript in their Reference section the first being a facsimile published in 1984 by publisher Frankfurt : Ma'had Tarikh al Ulum al Arabiya wa-al-Islamiyah reproduced from the MS Nuruosmaniye 3653, Istanbul and the second appears in "La Musique Arabe" by Baron Rudolphe von Erlanger originally published in Paris by P. Geuthner in 1930. These two books can only be referenced by visiting the British Library Reading Rooms - they cannot be obtained on loan or electronically copied due to copyright restrictions. So if anyone has access to these modern titles or is able to read them through a National Library it would be interesting to know if either volumes contain the oud engraving.
"La Musique Arabe" - 6 volumes, is currently being offered for sale via Addall Books for around $300 plus shipping - unfortunately a bit too expensive for me to purchase.

Thanks for the links suz_i_dil. All potentially useful information is welcome at this stage of the project.
Before picking up the thread again on construction of the oud, this might be a good time to pose two questions that sooner or later need to be resolved. The first is how to interpret the strange pegbox terminal on the engraving. From the limited pictorial evidence so far it might be concluded that the pegbox in the engraving represents a side view and that the 'finial' is, therefore. a flat plate of some shape. My guess is that it might be a 'tear drop' shape that can be found on other early oud like instruments as well as - oddly enough - the Portuguese guitar.
All other data that might support (or deny!) this speculation is welcome.

The second question concerns the number of sound holes. The engraving only represents two small sound holes but there is a possibility that an additional large sound hole may not have been shown because of the difficulty of engraving the sound hole as well as the strings. Also, I have not seen any pictures of early ouds with only two small sound holes but have to admit that I have not seen many pictures of early ouds from which to judge.
Peyman has noted that an early post Sassanid era (9th C) representation of an oud like instrument, chased on a metal artifact, has - what seem to be - two 'S' shaped sound holes. But are they? The 13th C Spanish 'Cantigas de Santa Maria" shows two players holding oud like instruments that have similar shaped 'sound holes' (also with what seems to be central sound holes?). Another illumination from the same book shows an oud like instrument with a triple rosette soundhole as well as two 'S' shaped features. Perhaps the latter are not sound holes at all but simply some kind of decorative inlay?
My current opinion is that I should build this oud with three soundholes - one large and two small. Alternatively, I could build it first with two small sound holes - exactly as represented in the engraving - and then remove the sound board to add a large central sound hole so that the two configurations might be tested and compared.
Again, all data that might help resolve this question is welcome.

Moving on to the next stage in construction - the mold bulkheads have been laid out geometrically (on both sides of each bulkhead) and cut roughly oversize on the bandsaw. Although I have used plywood for the mold baseplate I prefer to use pine for the bulkheads as it is a lot easier to work accurately than plywood. However, as pine is subject to shrinkage on drying, the pine boards (stored in an outside building) have been left to dry in a warm kitchen for several days before laying out the bulkheads. The metal template, for convenience, was used for the layout and the geometry then double checked using dividers to correct any inevitable slight deviations (pine being soft, marks made through the template with a scriber point tend to drift slightly when contacting the harder wood grain).

Now that the bulkhead geometry has been defined as accurately as possible, the rib profiles on each bulkhead are cut using a stiff backed razor saw (Lee Valley cat# 60F03.12 - good value for around $10). The saw cuts will remain as a reference after the bulkheads are finally profiled and shaped.

exellent! now we can see this thing shape up.

I have had less and less time to work on my project

hopefully I will find sometime before the summer.

The next stage is to trim the mold bulkhead profiles down to the circumferential scribed lines. If the bulkheads were made from plywood the shaping would best be done with a 'Surform' tool (for rough profiling) and finished with a fine wood rasp file followed by a sandpaper block. For pine bulkheads the most efficient method is to carve away most of the waste material with a paring chisel and then finish the job with a small block plane. Needless to say both chisel and plane need to be razor sharp to produce a clean, accurate finish.

To prevent chipping of the edges of the saw cuts as the bulkheads are shaped to their final dimensions, thin black card has been glued into the cuts. These inlays will also provide a clearer definition of the rib joint locations on the completed mold.

It is important that each bulk head should be mounted 'square' to the baseplate of the mold. Squaring each bulk head is easily accomplished by 'shooting' the bottom of each one using a shooting board and jointer plane. A small allowance of surplus material has been left on the bottom of each bulk head for this final trimming operation.

More pictures

As work progresses on the mold construction, work has started on the neck block.
The neck will be fitted to the neck block with a dovetail joint as is still the practice today so was probably the ancient method as well. The dimensions of the dovetail have been copied from an old Egyptian oud that I am currently restoring.
The neck block will be made from Sitka spruce that has been in stock for about 30 years so it should be well seasoned. Initially a block was laid out copying the same grain direction as that of the Egyptian oud - that is with the growth rings running vertically through the block. However, as spruce tends to split more easily radially across the growth rings (rather than along the growth ring direction) a second block has been cut with the growth ring direction running parallel to the top face of the block. My concern is that the first block may be more prone to splitting - a fault that can be seen in the Egyptian oud where the neck block has cracked across at the base of the dovetail. The second block should be less prone to this kind of failure. However, a sample of the spruce was split both along and across the growth ring direction with a chisel and there did not seem to be any noticeable difference in ease of splitting either way.
Nevertheless, on balance, the second block would seem to be the better choice although there is likely not a big difference.

The final step in shaping the mold bulkheads is to cut the 'flats' for the rib locations. These are cut using a wide paring chisel - to remove most of the waste - followed by a small block plane to make each 'flat' level and at the correct angles.
It is important to have good incident lighting for accurate work- progress in judging the material to be removed being aided by the light reflected off the freshly cut surfaces.

This oud will have an end plate - as dictated by the geometry - rather than the more common end block found on modern ouds (and the Arnault de Zwolle lute). This is a feature invariably found on surviving European lutes but can also (rarely?) be found on old ouds - as, for example, in this Egyptian oud. The end plate is fitted after the bowl has been removed from the mold so provision must be made for temporarily anchoring the ends of the ribs during assembly on the mold. Here a small block of pine has been shaped to the correct profile (using a metal template) and will be screwed to the mold baseplate as part of the bulkhead system.
The mold is now nearing completion with some small adjustments first to be made to ensure accurate alignment and levelling of the bulkheads on the baseplate.

Just for information and to illustrate different tail block/end plate configurations, image A shows the Arnault de Zwolle end block in section - similar to a modern oud but deeper. Image B - drawn to the same scale - shows the end plate, in section, of an early 17th C European lute and image C shows the end plate, in section, of the Egyptian oud previously mentioned. Note that the end plate of the Egyptian oud does not extend as far as the soundboard (it always does in early lutes) so to compensate for the lack of support in this area below the bridge, there is no half banding around the edge of the sound board over the end plate - only the full sound board thickness. As well, the Egyptian oud has a below bridge bar (invariably found on ouds?) that is located approximately at the inside edge position of the Arnault tail block. Clearly, sound board stiffness in the area below the bridge - no matter how it is achieved - is important.
In this oud reconstruction, the end plate will extend to the soundboard and will be of sufficient thickness to fully support the sound board in the area below the bridge - i.e. it will provide a surface area wider than the sound board half banding. There will also be a below bridge bar to provide the necessary additional soundboard support.

In order to layout the rib profile as precisely as possible a simple jig has been made. The bowl of the oud is semi circular in section so all of the ribs are identical in geometry (except for the two outside ribs that are to be made a little wider than the remainder and then trimmed to size).
The jig is made from two side plates of 3 mm thick hardboard - the required outline being traced from the mold base plate. The two plates have been joined with a cloth strip glued with flexible fabric glue (available from craft shops) so that they are hinged. The joined side plates were then smoothed down to match the exact profile of the base plate.
Several pine spacers have been cut to the required angle (using a metal template made from the bulkhead layout template) and glued between the side pieces - after checking that the gap between the side plates at each bulkhead location is correct (using dividers).
The correct rib profile is given by the inside edges of the jig and represents the inner profile of each rib (the finished outside dimension )will be a little larger.

After using the jig for making a rib template, additional wedges will be fitted and finished flush with outside edges of the side plates. This will then be used as a convenient jig for achieving the required profile of each rib during the bending operation.

To create the rib profile, a strip of paper is taped in place over the jig and rubbed against the edges of the jig with a finger. A clearer definition of the profile may be achieved using a little graphite powder (from a pencil lead).
The paper strip is then carefully removed from the jig, taped to a strip of card and the profile cut using scissors.
The card rib outline - with centre line marked - is then placed back on the jig to check for accuracy (paper and card are flexible so absolute accuracy cannot otherwise be guaranteed). In this case one side of the card template was found to be more accurate than the other so has been chosen for making a permanent rib template in thin sheet metal.

This template will give a rib profile that is slightly oversize so is just a convenient guide. Precise jointing of the ribs (with a plane) will be done during the final bowl assembly.

The neck block is to have a dovetail joint so this will be cut before the neck block is cut to shape - it will be a lot easier done now than later. As the centre line location of the block will be lost once the dovetail is cut, the block is first mounted on the mold baseboard with screws and register marks made with a knife so that the completed neck block may be later refitted in the correct alignment.

Interestingly, the neck block thickness, derived from the geometrical layout, is 1.5 inches (38 mm) which conforms nicely with modern day oud construction practice.

I have decided to use neck block #1 after all as the grain direction of #2 does not seem quite right.
After removing the block from the base plate, the waste material in the dovetail was removed by successive cuts on a bandsaw.
The sides of the dovetail were then trimmed to size with a chisel and finished with files - the end grain of the spruce being difficult to work cleanly even with a sharp chisel. The three sides of the joint were checked for straightness and squareness to the neck joint face with small engineer's square.
Finally the contour of the block was roughly cut on a bandsaw with the table set at 45 degrees. Ready for final carving to profile - but not today.

The completed bulkheads have been attached to the base plate using drywall screws that are thin so can be screwed into the soft pine without need for pilot holes. The bulkheads will not be glued just in case any ribs become inadvertently glued to the mold during assembly. The mold can be easily dismantled to remove it piece by piece if that occurs.
The alignment of the bulkhead centre lines was verified with a 'chalk line'.
All that needs to be done now is to shape the neck block in preparation for the bowl construction process.

Rough shaping of the neck block is being done using a sharp paring chisel and a metal template of the required finished curvature as a guide. The main problem is holding the block firmly while carving. I hold the work in the vise or clamped to the bench and take light paring cuts with the chisel. Pretty time consuming and a way to go yet before it is finished.

The neck cannot be finally fitted, aligned and shaped until the bowl is complete and the sound board fitted (but not glued) in place. However, the neck blank can be prepared and the dovetail tenon cut close to final dimensions - this work must be done sooner or later so I will go through this process now for information. The neck blank will then be stored in a heated room to ensure that it is completely dry.

The neck is to be made from a piece of Sitka Spruce. This stock was purchased, kiln dried, about 25 years ago. The grain of the wood is cut on the quarter - grain vertical - for stability of the neck - with minimum run out of the longitudinal grain.
The back of the neck will be veneered to protect it against damage from the fret knots when the frets are tied and slid into place.

I have a quantity of Lilac tree wood that is very hard (it is good for making pegs) so will plan to use that for veneering the back of the neck. Interestingly the heartwood is a pale lilac in colour (with white sap wood). I cut this about 25 years ago so it should be well seasoned.

To allow for later adjustment and trimming to size of the neck, the neck blank has been cut well oversize and longer than required.

The blank is first prepared by planing the upper surface (the 'face') flat and straight. Using the face as a reference, one side (the edge) is planed flat and straight and at right angles to the face. The angle is checked with an accurate engineer's square. Both face and edge are marked with pencil for reference. The opposite edge was then planed at right angles to the face and parallel to the reference edge (this is not absolutely essential, just nice to have, as all measurements for layout will only be made from the two marked reference surfaces). The back face of the blank has been left with a sawn finish.

The centre line of the blank is next marked with a marking gauge and highlighted with sharp pencil point run in the 'groove' left by the gauge.

Accurate preparation of the blank is important.

Finally, the dovetail tenon is marked out using a square and marking gauge - the lines all being cut with a sharp knife and filled in with a pencil for clarity.
The waste material is cut out with a hand saw leaving a small amount of surplus for final clean up and finishing.
Here, only the face of the neck joint has been finished. A wide paring chisel is used, freshly sharpened to a razor edge, to cut cleanly across the end grain. The knife cut layout marks serve as an indicator to guide the precise removal of material for accurate finishing.

One promising Ash timber examined measures 1.9 inches thick by 7.75 inches wide (48 mm X 197 mm) rough sawn and about 12 ft (3.7 metres) long. It had been sawn with the heart 'boxed in' so is mostly quarter sawn. The growth rings vary in width from under 1 mm near the centre to about 2 mm at the outside edges.
A growth ring count indicates that the tree was at least 150 years old when felled but was likely much older..
The density of the wood, calculated by measuring the volume and weight of a sample, is 56 grams/c.c. i.e. the specific gravity is about 0.56
The attached images show slices cut from each end of the timber to show the end grain. It can be seen that the heart runs out to one side over the length of the timber.

A macro examination of the cell structure (at the upper right hand edge of the top sample in the image previously posted) indicates that the species is Black Ash (Fraxinus nigra).

Selected sections of the timber will be cut into rib blanks tomorrow. One problem with reclaimed timbers like old floor joists is that they may contain hidden nails. The old iron nails have been removed but fragments of nail may still remain. The selected sections will, therefore, be cut longitudinally - to reveal the location and depth of old nail holes - using a carbide tipped saw blade before being re-sawn into rib blanks on a band saw.

The ribs of the bowl will be made about 1.5 mm thick so the bowl will be faceted or slightly fluted - like a 16th C lute bowl - rather than finished to a smooth exterior like a modern oud. Early Arabic/Persian texts indicate that the wood of an oud bowl should be thinner than the sound board.

The bowl could be made with a smooth exterior, with ribs finished to 1.5 mm thickness, if thicker ribs were used and the interior of the bowl was then thinned down by scraping after assembly. However, the rib blanks would initially need to be about 4.5 mm (3/16 inch) in thickness for a bowl of this size with 11 ribs. Not an impossible proposition but a lot more work. Perhaps some other time!

The neck block has been carved close to the required profile using a paring chisel and template as a guide to the correct shape. To finish the neck block, it has been mounted on the mold base plate and filed to size with a small wood rasp. The rib positions have been marked with pencil and the areas between flattened and smoothed with a file - using the pencil marks as guides. The mold is now 'ready to go'.

The sound board will be made from Sitka Spruce. I have a good supply of well seasoned material in stock to choose from. Sitka spruce is not an authentic choice like the denser softwoods - Larch and Cedar of Lebanon. A species of Larch is available locally but is unlikely to be of the quality required for instruments. Cedar of Lebanon is not an option unfortunately.

The sound board will be made from two, book-matched, pieces cut exactly on the quarter. Grain spacing is fairly wide for increased cross grain stiffness - to compensate for the lower density of spruce compared to Larch and Cedar of Lebanon.
The first step is to trim what will be the sound board joint with a straight edge and knife so that the grain runs parallel to the joint.

A sample of the selected sound board wood has been measured - its volume calculated - and weighed. The density is 0.43 grams/c.c. - heavy for Sitka Spruce but lighter than Larch or Cedar of Lebanon both with average densities of about 0.53 grams/c.c.

Before continuing with work on the sound board, this might be a good time to take a break and review aspects of the possible sound hole design. The oud engraving shows only two sound holes but it is possible that the engraver may not have represented a third large diameter sound hole because of the difficulty of representing both the strings and sound hole.
The sound holes may have been made in three possible ways :
1 - a hole cut through the soundboard with a separate decorative rosette (shamsa) glued behind the sound hole - as on a modern oud.
2 - a hole cut through the sound board and inlaid with a separate decorative rosette. This style of soundhole is found on early oud/lute related instruments like the gittern, vihuela and viola da mano.
3 - a decorative rosette cut directly into the sound board - the style of rosette invariably found on surviving lutes of the 16th and 17th C. 'Cut in' rosettes may ,therefore, have been adopted originally from early oud design (two oud like instruments depicted in the 13th C 'Cantigas de Santa Maria' have what appear to be 'cut' in rosettes).

It is not possible to determine from the oud engraving which of these three possibilities might apply. It has been decided, therefore, to 'cut in' the sound hole like a lute - this giving the best continuity and strength of the sound board structure in the area just above the bridge. I am experienced in making 'cut in' rosettes so the procedure might be of general interest.

The oud engraving represents a six point star motif for the rosette - most likely a greatly over simplified design for clarity. It has been suggested that the six point star design could imply a Jewish connection - perhaps the engraver himself was Jewish. However, I am more inclined to believe that the six point star in the rosette represent a basic element of traditional Arabic design.

The attached image shows a portion of a beautiful pierced window screen from the 14th C Mosque of Qawsun. Careful observation reveals the six point star motif appearing many times in the simple (yet complex!) design.
Compare this to two lute rosette designs - one from an early 16th C lute by Laux Maler and the other from the second half of the 16th C on a lute by Georg Gerle. The similarity to the window screen design is striking and so, strongly implies an Arabic origin. By the 16th C, these lute rosette designs were widely used 'standard' patterns and appear on many surviving lutes of the 16th and 17th C. so may also have been found on ouds and lutes from an earlier period.
To make smaller diameter rosettes - a lute maker would not scale down the design but would simply cut only that part of the original pattern covered by the reduced diameter. So, for example, the Maler rosette was likely made this way from a larger pattern (note that the points of the six point star in the design are 'cut off').

The Gerle pattern (conveniently, a favourite of mine!) is full size so will be used in this project - cut down for the small sound holes. If a large sound hole is also to be included then - interestingly - the pattern would fit the full diameter exactly.
Perhaps two sound boards should be made for testing - a 'two holer', as represented in the engraving and a 'three holer' as implied by the engraving. Perhaps even - two ouds should be constructed to enable direct comparisons to be made - all in the interests of science, music and history?!

....... and still pondering the enigmatic sickle shaped pegbox and spike shaped pegbox terminal of the oud engraving, here are some images from the 'Kitab al-mawalid' (the book of Nativities) by Abu Ma'shar al-Balkhi - taken from a 15th C copy translated into Arabic from the original 8th C Persian work.

The illuminations from the book can be seen in a slide sequence of images at
http://www.antiqillum.org/slideshows/Nativities/index.htm

The oud like instruments represented also show two-tone coloured sound boards (some with small sound holes?) and a smooth transition between neck and sound board - as discussed earlier in this thread.

The Spruce sound board panels have been set aside in a heated kitchen for final drying, before they are jointed and glued, so work continues, in the meantime, on the rib blanks.

The jig that was made to create the rib profile has now done its job so - with the addition of a few extra 'V' blocks and a pair of stands - will now serve as a jig (to determine the required rib curvature) during the bending operation. Not essential but a bit more convenient than using the mold for the same purpose.

The rib profile has been transferred from card to a metal pattern. This pattern is useful in determining the best part of the rib blanks to use (grain direction and figure) -while, at the same time, avoiding any flaws.
As the rib blanks have been cut in sequence from the billet, the ribs will each be laid out from a common centre - in this case the widest point of the rib - so that the grain and figuring of each rib will match when assembled into the bowl.

On closer inspection, several of the sawn rib blanks cannot be used because of the location of scars left by the old nail holes. It has been decided, therefore, to use two, now incomplete, sets of ribs - arranging the ribs alternately in sequence. There are 8 good rib blanks in one set and 7 in the other so there is still plenty of room to maneuver in the final choice of ribs.
One rib has a pin knot running across the width of the blank. It may still be possible to use this if the fault (a structural weakness) can be located inside the rib and over the neck block.

The saw marks on each rib blank have been removed with a smoothing plane and the ribs brought down to about 2 mm thickness. The plane does not have an adjustable throat so - even with the relatively plain grain figuring of the wood and working the plane at an angle - some slight 'tear out' of the grain is inevitable. Final smoothing and thicknessing (to about 1.5 mm) will, therefore, be accomplished using a cabinet scraper.

The batch of sound board material to be used on this project was cut many years ago at a local sawmill from a billet of Sitka spruce - measuring 8 inches X 4 inches X 4 ft long
(20 cm X 10 cm X 120 cm). (The Spruce was originally purchased, during the 1970's, from a company that made gliders - used for making airframes. The company (the Kirby Glider Co. of Yorkshire, U.K.) kindly allowed me to select the only two timbers from their entire stock that I judged would be suitable for instrument sound boards. Great customer service!).
Unfortunately, the local sawyer ran this billet through his bandsaw using power feed rather than by carefully hand feeding the work (this was a large bandsaw - with blade about 6 inches (15 cm) wide - used for resawing large timbers). Consequently, the batch of soundboard blanks came out of the saw straight on one side but with a pronounced bow on the other. Rather than scrapping the whole batch, each sawn blank was cut down the middle to eliminate the bowed portions. This resulted in a number of blanks about 4 inches (10 cm) wide up to 30 inches (76 cm) long with no longitudinal twist and with only a slight longitudinal 'ripple' or 'waviness'(this will eventually be planed out when each sound board is reduced to its required thickness).
These blanks will make four sound boards (each made from four pieces - the two piece soundboard mentioned in an earlier posting will not now be used). The pieces will not be 'book matched', in the conventional way, but will be selected and joined so that any longitudinal ripples will exactly match. This is to eliminate any potential 'built in' stresses in the finished soundboards.
I might as well make up the four soundboards at one time and choose the best for this project.

The first step is to 'shoot' the edges of each blank so that the longitudinal grain runs parallel to the joint edge.

I am attempting on this project to salvage sound board material that would otherwise have been thrown out. The panels have all been cut in sequence from a billet with end grain cut perfectly on the quarter and are well seasoned so should potentially make reasonably satisfactory sound boards.
All of the panels - because of the way they were sawn - have a slight but matching longitudinal bow or curve. To compensate for this in the completed sound board, the panels will not be book matched as any longitudinal 'bow' will be reversed along the central glue line that could result in built in stress. On the contrary, if the panels are not book matched, any longitudinal bow will be perfectly matched between the panels - eliminating any danger of built in stress from that cause. The panels will not be switched end for end.
Is this approach likely to be significant? Hard to tell without trying. This is an experimental project so I have an excuse to try to find out!

This approach might also be beneficial - for the same reasoning - in the case of sound board panels with longitudinal grain run out. This is caused by spiral growth commonly occurring in trees and so is likely to be found - to a greater of lesser degree - in a lot of sound board material offered for sale on the open market?

Note that many early ouds (and lutes) had their sound boards made from several (three to five) - sometimes mismatching - pieces rather than from two 'book matched' panels as is the common practice today.

The attached sketches should clarify what I am trying to explain.

Note that this thread is not intended to be a tutorial claiming to be the "right" way or the "best" way to do things but is just an ongoing record of my approach (rightly or wrongly) to creating this instrument.

The sound board panels, having been rough planed to remove the saw marks, have now been sorted into four sets sets of four panels ready to be jointed. It is helpful to use a half template of the soundboard profile to help visualise and determine the best positioning of the panels. All of the panels show a pronounced "cross silk" figuring - a good sign indicating that the grain is cut perfectly on the quarter. The colourful grain doesn't bother me!
I will be able to make up four soundboards - perhaps five - from this batch of wood.

Two sets of panels (to make two complete sound boards) have been jointed ready for gluing. The jointing process is quite critical as the joint surfaces must be planed accurately enough to be practically invisible (viewed on both sides of the panels) without forcing the joint together. I judge the precision of the joint by eye as well as by feel - flexing the joint to ensure rubbing contact along its length. The joint surface is left as it comes off the plane as the smooth surface results in the strongest glued joint. Not having built an instrument for a few years now, it took me a little while to get the hang of jointing sound boards again (frustrating!) - a good reason to make up a batch of sound boards at one time.

A batch of hide glue has been made up and is currently soaking ready for use tomorrow in gluing up the sound board.

Taking another interlude from the practical aspects of this thread to consider another design detail - the bridge.

The representation of the bridge in the engraving of the oud is distinctly 'lute like'. As no ouds survive from earlier than the late 18th C. it may be useful to look to surviving lutes for clues as to how ouds once may have been constructed.

The decorative carved ends of the bridge are reminiscent of two unusual examples found on surviving lutes. The first is to be found on a lute by Magno Dieffopruchar, Venice, 1612 cat. 1753 in the Museo Civico, Bolognia and the second on a lute by Christofolo Choc, Venice 1630, cat. 7756-1862, Victoria and Albert museum, London. The attached images show a tracing of the bridge end of the former and a picture of the bridge end of a copy of the latter currently under construction (see "Old Project - New Lute" on this forum).
The bridge in the engraving would appear to have tile inlays on the top front and rear edges of the bridge - presumably to protect the edges of the bridge against wear from the strings. The main body of the bridge would appear to be 'in the white' or unstained (compared to the tuning pegs in the engraving that are represented in black).

With these clues in mind, the bridge of the reproduction oud will be made from pear wood left unstained in its natural colour with tile inlays of African Ebony and Persian Boxwood. The ends of the bridge will be carved similar to the lute designs.
A matching tile inlay will be repeated around the edge of the sound board - the half binding (or banding) is represented in the engraving but without details of the inlay pattern.

Thanks Peyman.

Relative humidity in the heated kitchen is currently around 45% so today was a good time to make a start on gluing up the soundboard panels.
The work surface is made from two pieces of 15 mm thick MDF (medium density fibreboard) glued together. This makes a firm, flat and stable surface.
The gluing technique is well known - standard methodology, not 'rocket science! Two panels are laid edge to edge on a strip of wood (about 10 mm thick X 20 mm wide in section) so that the edges to be glued are temporarily slightly elevated. The panels are held firmly in this position with pins driven into the work surface. This arrangement is to provide a clamping force on the glued joint when the panels are pressed into place against the work surface.
To prevent the panels sticking to the work surface, a strip of 'non stick' cooking paper is laid underneath the joint. Hot glue is then applied along the joint surface, the strip of wood pulled from underneath and the joint pressed to the work surface along its full length. Another strip of 'non stick' paper is then placed over the joint and the panels held in place with a wooden batten clamped at each end.
The non stick paper works quite well but a strip of newspaper can also be used as an alternative.
One (of many) advantage of using hot hide glue is that clamping time is relatively short so the panels may be removed from the work surface after about an hour - allowing another set to be glued up. The glued panels removed from the work surface will be allowed to dry fully overnight before they - in turn - will be glued together to complete the soundboard.