Old Oud - New Project

The banding tiles have been glued to the prepared fingerboard blank and the assembly planed close to finished dimensions.

To plane the fingerboard to the required depth (4 mm thick at the nut end and 2mm at the neck joint), the assembly was stuck to a flat board with double sided adhesive carpet tape. The brand of fabric tape that I am using has quite a strong adhesive - more than adequate to hold the work during planing. As the adhesive softens at a fairly low temperature, the fingerboard was released by applying heat with a hair drier. Unfortunately, although the tape released cleanly from the board, quite a bit of the sticky adhesive remained on the underside of the fingerboard. Most of this was scraped away while soft and the residue was cleaned up with a gum solvent ('Goo-Gone') followed by a cleaning with isopropyl alcohol.

To restore the 'missing' 1 mm. 'set back' of the neck (most likely caused by the problem encountered earlier with gluing the neck joint), it may be necessary to plane a little from the neck - if sufficient material remains to allow the consequential reduction of the width of the neck at the nut. If not, the fingerboard will be planed down in thickness at the nut end as necessary to set the required action.

Once glued in place, the fingerboard will be finished to its final contours. As this is a fretted instrument, the fingerboard will be given a slight convex curve across its width (i.e. 'crowned'). This is necessary to ensure that the 'tied on' gut frets fit tightly to the fingerboard surface and do not cause string 'buzzing'. This slight curvature must also be allowed for in setting the string action.

With the slightly oversize assembled fingerboard now glued to the neck it has been trimmed to fit on the edges by filing and scraping.
The fingerboard surface has been brought to the level of the sound board at the neck joint and finished flat with a cabinet scraper. The scraper removes very fine shavings necessary for this operation.
A small amount needs to be removed from the nut end. This will be done when the nut is fitted, the fingerboard 'crowned' and the action finally set.
The 'crowning' is assymetric being deeper on the treble side than on the bass side (to keep the action as low as possible on the treble side). The target will be to have a string clearance at the neck joint of 2.5 to 3 mm on the treble side and about 4 mm on the bass side - additional clearance being required for the thicker (silk) bass strings and their greater amplitude of vibration. Also, for this reason, the depth of the bridge tapers towards the treble side.

The attached image shows the assembled oud so far. Slowly 'getting there' but still quite a bit of work to be done.

The next step is to cut the rebate for the sound board binding, cut the tiles and glue them in place - which should take a while to complete!
The material for the tiles (Ebony and Persian boxwood) has been prepared by planing strips to a thickness of just over
1 mm. These will be further cut into 5 mm wide (oversize) strips for making the individual tiles - using a cutting jig to ensure consistent size and angle of cut.

A tile cutting jig is an easily made tool.
A 5 mm wide slot has been milled in a piece of hardwood to a depth of about 5 mm to retain the veneer strip during cutting.
The tiles are wedge shaped, 5 mm wide on the inside edge with an included angle of 2 degrees for the tiles around the semi circular lower part of the sound board and 1 degree for the upper part of the sound board that has a radius of curvature double that of the lower part. These angles were determined by drawing a full sized layout plan.
Although small, the angles are (theoretically) significant cumulatively when the tiles are joined together - although, in practice, slight joint mismatch between the tiles may be largely disguised by the contrasting colours of Ebony and Boxwood and the dark hide glue line.
In order to measure the angles with some reasonable level of accuracy, a low cost angle measuring tool was used that has a vernier device that allows measurement (theoretically) to a tenth of a degree.
With the correct tile side angles marked on the jig with a knife, slots were cut with a fine razor saw to just beyond the depth of the milled slot. The saw blade was guided vertically during the cut by using a square edged piece of metal clamped to the jig.

The prepared veneer strip was trimmed in width to fit the slot in the jig using a mini shooting board - a 5 mm slot, 2 mm deep milled in the edge of a board with a router.

In use, the end each veneer strip is first cut to the required angle in the jig using the razor saw (a couple of draw strokes is sufficient to cut through the veneer). A thin piece of metal is then inserted into the saw cut to act as an end stop. The veneer strip is then reversed, pushed up against the stop and the second edge of the tile cut with the saw. The veneer strip is again reversed and the process repeated to produce the next tile.
Seven tiles have been cut - only another 230 to go!

The first few tiles will be used to test the accuracy and viability of the method before going into 'mass production'.

In order to test the effectiveness of the tile cutting jig, a sample of sound board was first prepared from a piece of scrap Sitka spruce, cut to the contour of the lower part of the sound board (16.8 cm radius).
A half banding rebate, 4 mm wide and about 0.8 mm deep, was then cut by first scribing the inner radius with a purfling cutter followed by a double sided rebate plane - both 'home built' special purpose tools.

If square tiles are used for the sound board edge banding - each individually glued in place - there will be a gap between the tiles on the outside edge of the sound board. If the tiles are set correctly at an equal 2 degree angle to each other, the gap will measure 0.14 mm (or 0.006 inch) - see image A. However, in practice, the gap between the tiles might vary between zero and 0.23 mm ( 0.009 inch) maximum if the tiles are not set at precisely the correct angle - see
image B . (the sketch is drawn with exaggerated proportions for clarity)

The tile cutting jig is designed to eliminate the gap between the tiles by cutting each tile to a wedge shape with an included angle of 2 degrees (for tiles in the bottom section of the sound board). However, in practice, it has been found that - with the razor saw kerf measuring about 0.010 inches wide (0.25 mm) together with working clearances and other slight cumulative dimensional errors in making the jig - that the tiles cannot be cut with sufficient accuracy to maintain the required wedge shape of 2 degrees included angle. The test samples all measured about half that degree of taper or less. Furthermore the saw cut edges introduce further lack of precision.

Comparing a set of square cut tiles with those cut on the jig, the attached image shows that there is little measurable difference in the fit between the tiles (the jig cut tiles are those with the pencil marks indicating the outside or widest dimension). The tiles are oversize by about 1 mm so much of the visible gap will be removed when the sound board edge is trimmed to finished dimensions. Also, any gap between the tiles will be filled by the dark coloured hide glue used to attach the tiles to the rebate so should be practically invisible.
For tiles on the upper part of the sound board the gap will be half that of the lower tiles.

It has, therefore, been decided to move ahead with square section tiles - these being easier to cut accurately than the wedge shaped tiles. The tile cutting jig has been modified to make the square cuts from straight edged 5 mm wide veneer strips.

To complete trials with the banding tiles, both sets of tiles - square cut and taper cut have been glued to the sound board test piece. The purpose of this test was to determine the best method of gluing the tiles to the sound board edge rebate and to evaluate the best tile 'cut' to use.

As the glue quantity requirement is very small for this operation, about 15 ml of 'pearl' glue were heated in a 30 ml plastic mixing cup ($4.40 per 100 from Lee Valley). Pearl glue was chosen as it is a darker colour than the higher strength hide glue that I use for instrument work. Water temperature was maintained at around 140 Fahrenheit with a dial thermometer.

Hide glue gels quickly so I was prepared to have to reheat the glue with a hot iron at some time during the procedure to ensure all of the tiles had 'bedded' properly in the rebate. This was done by laying paper strips (cut from the print free edges of a newspaper) over the tiles and applying the hot iron to remelt the glue and press the tiles firmly in contact with the rebate.
However, this part of the process was judged to be unnecessary as it was found to be quite easy to push each tile exactly into place - the tile then being almost immediately held in position by the rapid gelling of the glue.
Nevertheless, in the event of a tile not being seated properly, the hot iron technique is the way to correct the situation.

Otherwise the gluing of the tiles was found to be a quick and uncomplicated process - if not a little messy (a damp cloth to remove gelled glue from fingertips as work progresses is probably a good idea).

The test piece was finished by scraping away surplus glue (already hardened) and material about half an hour after the tiles had been glued in place. (Normally a full day should be left for the glue to dry).

Macro images were used to assess the results - the digital camera having a far better resolution than my eyes! To judge the scale, tile width is 5 mm and finished depth about 0.6 mm.
- and overall thickness at the edge of the sound board test piece is 1.4 mm. (I was a bit too'heavy handed' finishing with the scraper).
On the oud, the depth of the rebate will be increased slightly to give a finished tile thickness of about 0.8 mm.

On balance, the taper cut tiles - despite the tolerance inaccuracies inherent in the tile cutting jig - seem to be a cleaner, more precise fit than the square cut tiles.

No question - use of hot hide glue is the simplest and most efficient way to go for gluing half banding tiles!

Work continued today on the sound board edge half banding by cutting the rebate.
The rebate inside edge is first scribed to the depth of the rebate (about 0.8 mm) using a purfling cutter. This is one that I made from brass some years ago that does a reasonable job. The spear shaped blade tip (flat on the outside edge - i.e. the inside edge of the rebate) allows the tool to cut in either direction. The width of cut is set to 4 mm using the two brass screws and the blade tip to project 0.8 mm (trial cuts on a piece of scrap wood are made to verify the settings).
The scribing cut is made with several light passes of the tool around the edge of the sound board until full depth of cut is achieved.
As the tool cannot quite reach into the neck joint this part will be finished with a knife and chisel.

The rebate is being cut with a purpose designed, double sided, miniature rebate plane (made, step by step as a topic on this forum a while ago).
The plane cuts a rebate of fixed width and depth. The design allows the plane to be used on either a 'pull' or 'push' stroke - dependent upon the wood grain direction of the sound board.

No room for error here so the rebate must be cut to depth with several passes - eventually setting the blade a little deeper for the final cut to the required depth. The most difficult part is cutting across the end grain at the bottom of the sound board to avoid tear out. This part will, therefore, be finished to depth with a 'safe edged' file.

I decided to take the cutting of the rebate slowly, carefully and leisurely, part done this afternoon for half an hour and - taking a break - finished tomorrow.

Unfortunately, "taking a break" has now taken on a new meaning!!

To hold the oud firmly during the operation of cutting the rebate, I used a pad of non-slip rubber carpet underlay on my lap - oud bowls being a difficult shape to hold without slipping and sliding.
However, at the end of the afternoon - when altering the position of the oud on my lap for a final cut with the rebate plane - the oud suddenly shot from my grasp landing on the floor - some distance away - with a loud 'crack'. The peg box had broken off with the force of the impact!
Fortunately, the hard floor of the room is covered with carpet so the damage appears to be limited only to the area around the peg box/neck joint. The hide glue withstood the test with the wood fracturing rather than the glue joint!
As this joint design replicates that on an old Egyptian oud in my possession, it is interesting to note that the fracture replicates almost exactly the peg box damage found on the old instrument. So, I guess that this 'break' may be considered to be historically correct!

The peg box will be repaired with a new end block. The repaired peg box will be re-glued to the neck after completion of the sound board edge banding. Rather than replace the whole back plate of the peg box, a 'beard shaped' patch of ebony will replace the broken area.
Just a bit more extra work.

Quote: Originally posted by jdowning

...I guess that this 'break' may be considered to be historically correct!

Nice to see you can still keep a sense of humor in such an unfortunate situation!

All the best for this fine project.

The repair work should not be much of a set back Sazi as it will be a relatively straightforward job. On further examination, the back plate is also split part way along its length so will be completely replaced as well. The new end block will be made from elm this time and not spruce.
The end block will be fitted and glued in place first (using the pegs already fitted to maintain correct alignment). Then the damaged back plate will be removed and replaced.
The wood stock for the repair is now rough cut and will 'season' for a few days in our heated kitchen before use.

Quite a bit 'warmer' today than it has been (just below zero degrees) so the banding rebate work was completed in my work shop - this time with the oud safely supported on the bench!
The rebate plane did a reasonable job dealing with the fairly wide grained Sitka Spruce sound board material. Any small irregularities in the the rebate were finished level using a Nicholson 'safe edged' auger bit file (Lee Valley cat# 62W08.01 @$11.90) - a good quality tool that might have been designed for this job.
The boxwood and ebony strips for the banding tiles have been prepared so the next task is to cut the tiles using the tile cutting jig. I shall need about 65 each of boxwood and ebony tiles for the upper part of the sound board edge and about 55 of each kind for the bottom part - say about 250 in total allowing for a few spares.

Rather than change the design/construction philosophy at this stage, I have decided to repair the peg box end block again using Spruce rather than Elm - as Elm will add little, if anything, to the overall strength of the joint, the neck core being made from Spruce (veneered with Ash).
Peg box to neck joints are relatively weak due to the proportion of wood end grain (at the end face of the neck joint and peg box sides) so best to avoid dropping an oud onto its peg box in the first place!
I shall make the end block a bit deeper than before for additional strength.

The first step in the repair is to fit and glue the replacement end block. The end block blank is made well oversize in order to make a precise fit of the slight taper easy to achieve. This done, the blank may then be trimmed close to size and glued in place. The peg box back plate at this point remains in position to maintain overall alignment.

The sound board edge tiles have been glued in place and will be left overnight for the glue to dry thoroughly before final trimming.

The tiles were first prepared, after cutting on the jig, by marking each tile with pencil to indicate correct orientation and to remove any 'burrs' left by the razor saw underneath each tile. This was a fairly time consuming but important step to ensure that the gluing operation could proceed quickly and smoothly.
The tiles were laid out in correct groups and orientation on a cheap kitchen plastic cutting board. Glue will not stick to the surface but - more importantly - the board is relatively heavy so that there is little danger of upsetting the tiles halfway through the job as might happen if the tiles were laid out on a sheet of paper - which would be a setback.

Hot hide 'Pearl' glue - maintained at a temperature of 140 F (60 C) - was used for gluing the tiles into the rebate. This grade of glue is not quite as strong as the Granular hide glue that I use for structural components like the ribs and braces but takes longer to 'gel' so is perfect for the tile gluing operation allowing time to correctly position each tile. Nevertheless, it is necessary to move forward with the gluing operation at a steady pace.
As a precaution, a hot iron was left 'on stand by' in the event that some tiles were found not to be seated properly in the rebate and so could be ironed in place. This was not found to be necessary however.

The edge of the sound board was marked (through the centre line of the two rosettes) to indicate the transition from square cut tiles to the wedge cut tiles around the bottom of the sound board.

The procedure for gluing was to use a small paint brush to apply a dab of glue to the rebate surface and then press each tile into position wiping away any surplus glue with thumb or forefinger of the left hand. Before picking up the next tile (using the left hand) the fingers were wiped on a damp cloth to remove any glue. The wet finger tips also make it easier to position the tiles as the tiles will not readily stick to the fingers.
No additional clamping or taping of the tiles is required.

The tiles were glued on one side of the sound board at a time (one side in the morning the other during the afternoon) giving time for the glue to harden sufficiently on the first side.

The last tile will have to be custom fitted as the remaining gap is just a little too tight to fit a tile without trimming. Also, the black/yellow tile colour sequence did not quite work out perfectly so the last tile will be black.

The tiles are oversize so about 1 mm will be trimmed around the edge of the sound board and about 0.25 mm removed to bring the tiles flush with the sound board surface.

As this project advances to completion (but not wanting to rush things!) there are a few items yet to consider ahead of time.
The fret gut has arrived direct from 'Pyramid Strings' in Germany (with the usual fast and efficient service).

I had hoped to purchase bone blanks for the nut from luthier supply companies in the USA and Canada but shipping and handling far outweighs the cost of the blanks. I may be able to obtain bone blanks from the nearest music shop - half an hours drive away - but then there is the cost of gas and time involved which again outweighs the value of the blanks (if they stock them).
I may try Persian boxwood as a nut material - cut on the end grain which is very hard - as I have plenty of that material.
However, as it may be of general interest, I shall try to make bone nut blanks 'from scratch'. So, a 'soup bone' (leg bone of a beef cow) was today purchased from a local butcher ($2.50) with the knuckle joints trimmed off to leave a relatively straight section of bone in between. I have no idea how to select the best bone for the job but must start somewhere. A brief search on the Internet indicates that there is much information on how best to undertake bone preparation for use in musical instrument work. So, with our dog showing great interest in the raw material, it goes into the refrigerator for now until I can decide how best to proceed. More to follow!

Concerning the finish to be applied to the oud and as this is an experimental project with none of the usual commercial restraints (if it doesn't work - no customer to satisfy so no big deal) I am curious to try traditional 'glue size' as a grain sealer for the Ash wood.
Some test pieces of Ash wood have been tested to confirm compatibility with possible finishes such as Shellac or oil varnish.
Glue size is just a dilute solution of hot hide glue in water. For this test the concentration was 25 ml of 'Pearl' glue (fluid) in 250 ml water. This was brushed on to the test pieces at 140 C.
Each test piece immediately expanded into a convex curvature as moisture was absorbed into the surface of application. For example a test piece 48 mm wide and 2 mm thick arched into a significant curve 2 mm deep in the centre. However, once the surface moisture had evaporated, shrinkage of the dried glue resulted in an almost equal and opposite concave curvature.
Glue size (as well as water) 'raises the grain or fibres' of the wood (an important part of the finishing process) - these raised fibres and 'whiskers' then being removed by scraping or fine sanding to produce a smooth surface in preparation for the final finish (Shellac or Oil Varnish).

The edge banding of the sound board has been levelled and finished to size.
To avoid breaking or chipping the tiles, a flat single cut file has been used to dress the tile edges to the profile of the sound board. (Only files and scrapers will do the job - slow but safe)
The same file was then used to roughly level all of the tiles to an equal height.
The tiles were then reduced to sound board thickness with a curved scraper blade followed, finally, by a flat scraper blade (with masking tape on the outer edges to avoid possible damage to the sound board surface). The curved scraper allows material and dried surplus glue to be removed locally close to the banding tiles without contacting the sound board surface. The scrapers are worked at an angle to the tiles (and sound board grain) to prevent the scraper 'catching' in the joints between each tile which might cause irregularity in the finished surface.

To complete the work, the sharp edges of the banding tiles have been rounded over with a single cut file. (I avoid using sand paper wherever possible when finishing wood)

A job not to be rushed - taking about 4 hours to complete.

The damaged pegbox has been repaired with a new end block and back plate. The backplate has been made a bit thicker than before for extra strength. After a 'dry run' to check the accuracy of the pegbox to neck joint, the peg box has been glued in place.

Before gluing the peg box in place, the finger board was finished level ( using a scraper and double cut flat file to remove any slight 'high spots'). As the oud will be fretted, the finger board surface is slightly 'crowned' to ensure that the tied on gut frets lie tight against the fingerboard.
It is interesting to note the comments about flat versus rounded fingerboards on lutes written in 1676 by Thomas Mace (Musick's Monument). Mace published his book in an attempt to regenerate interest in the lute that - in his time - was waning in popularity (eventually to die out completely by the mid 18th C).
He warns about the flat fingerboards found on lutes of the 16th C as being a fault. However, the lute of Mace's time had more strings and a much wider fingerboard than the five or six course lutes of the early 16th C - so rounding of the fingerboard surface of his lutes was essential to avoid string 'buzzing' or sound damping due to loose frets.
Nevertheless, even for the relatively narrow fingerboard of the oud of this project, the finger board surface has been very slightly 'crowned' and the edges of the finger board smoothly rounded.

With the sound board edge banding levelled and cleaned up, my maker's brand was used to mark the sound board just below the fingerboard. Branding a lute sound board was the practice of some 16th C lute makers - the brand sometimes attempting to disguise the wooden guide pin in that location used for sound board alignment.
The branding tool is easily made from brass strip bent to the required shape and hard soldered (brazed or silver soldered) to a metal rod. The fine detail is then done with needle files. All in reverse, of course!
In use, the brand is heated to scorching temperature with a propane torch - tested first on wood scrap - and then pressed in place to make the brand mark.

At this rate of progress it looks as though I am going to have to drive to the nearest town to purchase a bone nut blank after all.

Conditions in my kitchen are perfect for varnishing (25 C/72 F @ 45% Relative Humidity) so work on finishing the oud started yesterday with a first varnish sealing coat on bowl, neck and peg box (the sound board will be left unvarnished).

I have decided not to use glue size as a wood grain sealer when I realised that the grain had already been sealed with glue as a consequence of constructing the oud bowl using the ancient technique of applying glued paper strips to hold the rib joints together during assembly. Also, I was a bit nervous about the possibility of introducing stresses to the bowl - due to glue shrinkage - now that the sound board is in place.

I am using an oil varnish to finish the oud. In the past I have used violin oil varnishes to finish lute bowls - slow to dry and difficult to apply, particularly if the varnish is coloured with stain.
For this project I am using a so called 'hand rubbed' oil varnish - Minwax Antique Oil. I have used this product successfully on many wood working projects in the past but not for instrument making. The 'hand rubbed' varnishes are all basically a mixture of linseed oil and varnish. Linseed oil takes forever (never?) to dry so addition of the varnish speeds up the drying process to within a few days.
The Minwax product dries hard enough to re-work in 24 hours. Ash wood is an open grained wood so the method I am using aims to fill the grain (with the varnish) to produce a smooth surface (but not a glossy 'French Polish' finish). Most of the varnish coats will eventually be removed by sanding with fine sandpaper (320/400 grit - 3M 'non clog' paper is the best) leaving a thin, light, protective layer of varnish at the end.

The first sealing coat of varnish has been applied and allowed to dry overnight. This coat has been lightly rubbed down with 320 grit paper - to level any 'high spots'. This produces a fine varnish dust that will be used as a grain filler during the next finishing stages.
The varnish is applied with low cost, disposable 'foam brushes' (wearing plastic gloves for protection). Once the varnish has been allowed to 'cure' for about 5 minutes, surplus varnish is then wiped off with a piece of cotton cloth and the work allowed to dry.

All used foam brushes, cloths etc are a potential fire hazard until the varnish has hardened so are left outside - in the snow - for safety.

In the meantime!
Local music shops do not stock bone nut blanks.
Making my own own blanks from scratch using cow bone may take a few weeks to accomplish so thought that I would try using what I already have in stock.
None of the bone pieces are big enough for the nut so thought that it would be interesting to make a 'composite' nut - Ebony with bone inlay. Easily made - the ebony blank being rebated for the bone inlay. I shall glue the bone to the ebony with epoxy cement but a better choice would be the traditional luthier glue (wood to ivory joints) made from hide glue and isinglass (sturgeon glue).

Varnishing of the oud will take a week or so to complete as each layer of varnish must be allowed to dry before being rubbed down to the wood with steel wool. Three coats have been applied so far. It may take another four or five 'rubbed down' coats to finish the job.
In the meantime other work can proceed.
The Brazilian Rosewood pegs - refurbished from a set that I made years ago for a lute - have been 'inlaid' with pure tin/silver alloy wire (as reported in a previous post) a technique that I wanted to try out on this project.
The tin/silver wire is quite soft and easy to form and can be cut with a sharp knife. The split wire rings sit in a groove hand filed in the peg shank just below the peg head. Each ring must be trimmed in length so that it fits exactly in the groove when compressed into position.
The rings have been glued in place with epoxy resin for convenience although hide/isinglass glue or shellac would likely do just as well. Brazilian Rosewood is an oily wood - great for finishing to a polished surface but problematic for gluing. All surfaces must, therefore, be thoroughly de-greased before gluing. I have used 100% isopropyl alcohol. In the past have used de-greasing agents like 'Genklene' (used to remove grease spots from clothing) but these (hazardous) products may now be unavailable on the consumer market.

In addition to the 'silver' ring, a silver 'dot' has been inlaid into each peg head. Easily done by drilling a short hole of suitable diameter (1/16 inch in this case) in the peg head so that the wire is a tight fit in the hole. The wire is then lightly hammered into the hole (without need for glue) and trimmed flush with the peg surface.

All surplus cured glue has been removed with a 'safe edged' file and fine steel wool.

All that remains is to give each peg a final hand polishing with cloth strips/string (or perhaps with a felt buffing wheel in a 'Dremel' type tool).

Taking a break from varnishing the bowl, the nut rebate for the composite bone/ebony nut has been finished and the nut shaped and fitted.
To hold the oud securely for this work, a clamping jig was made from scrap materials - padded with rubber carpet underlay to prevent damage to the neck. The jig is conveniently held in the bench vise.
The nut rebate was finished square with a sanding block - easily made from a piece of pine, planed square with some 120 grit Garnet or sand paper glued to one face. Final adjustments to the rebate were then made using a mini scraper blade.

The composite nut is just sufficient in length to make two nuts - a useful spare if needed. The nut has been planed and filed square to precisely fit the rebate and roughly finished to the required contour. The nut will be finished just prior to stringing the oud.

A final coat of varnish has been applied to the bowl after rubbing down all of the previous varnish coats (5 coats I think but have lost count). The final sanding was done with 400 grit 3M non clog paper (good stuff) followed by 0000 grade steel wool. So, after all of this work, only one coat of varnish - the final coat - remains! I am not sure if this will be sufficient to fill all of the open wood grain but will wax the bowl which should fill any remaining pores to a smooth surface.
The neck will be varnished next followed by the peg box - both varnished separately just for ease of handling.
That should bring the project to the final stage of stringing, action adjustment, fretting etc. perhaps a week or so from now - all being well.

Hi John,
I hope to have a chance to see this beauty when I pass by sometime in the future.

Is it really going to have frets? if so, would the action of the string allow to play the instrument with the frets removed? or is it going to be a strictly fretted instrument? just wondering.

The final varnish coat was re-applied today but with similar negative results.

Hand rubbed oil varnish is applied to the surface of the wood, allowed to dry for 5 to 10 minutes, then surplus varnish is wiped off with a cloth until the surface loses any 'stickiness' after which is allowed to dry for 24 hours.

This procedure was followed but after about half an hour small spots of varnish started to appear on the finished surface - coming from the open pores of the wood. These were wiped away with a dry cloth to remove any 'stickiness' which seems to have corrected the problem - although some small specks of varnish still remain in places. So the problem seems to be related to the open grained Ash wood trapping uncured varnish.
Rather than attempting to completely fill the wood grain with repeated applications of varnish, this final coat will be buffed to smooth out any remaining small varnish spots and then waxed.

This kind of finish is not the same as a brushed on instrument varnish (as used on violins) - however it does help to 'bring out' the grain of the wood and provide a hard protective finish.
The Ash wood of the bowl is relatively plain (as far as woods go) but, nevertheless, has a some complex but subdued 'fiddle back' figuring.

Varnishing is now complete and the banding tiles cleaned of any residual varnish, along the front edges, using a scraper blade and file.

The peg holes have been cleaned of any residual varnish using a peg reamer (with a light touch to avoid removing any wood). The pegs have been fitted using dry soap as a lubricant - important at a time of year when humidity changes might cause the pegs to jam. The soap is just regular hard laundry soap that hasn't been near water for years so is powdery dry.

The nut has been fitted and the string grooves marked in pencil using a rubbing taken from the nut of one of my lutes. The grooves have been filed with needle files and burnished smooth with a piece of string. The string height has been left at just over 1 mm so may still require further adjustment.
String action and alignment has been checked on the treble and bass sides using thin fishing line as string. The action is currently about 0.5 mm higher than I want but will leave it at that for now until it is seen how the instrument 'moves' at full string tension. About 2 mm extra finger board thickness at the nut has been allowed for possible action lowering adjustments.
Target string clearances at the neck joint are 3 mm on the treble side and 4 mm bass side.

The completed oud (without strings and frets) weighs 624 grams - well within the 650 gram target weight.

Now to calculate the strings required. These will be Pyramid PVF with metal overspun basses for initial tests switching later to all silk strings. The oud will be strung first without frets so that the action can be verified and any adjustments made.

The oud has now been fitted with strings and frets for an initial assessment. The strings fitted are those that I have to hand, including plain nylon trebles - just to bring the instrument up to tension. These will later be replaced by the correct Pyramid PVF trebles. The frets are also temporary - 0.8 mm nylon fishing line (much cheaper than gut!) - just to get a feel for things.
Apart from fitting better strings, the action feels too high. Also the string pairs need to be a little closer. So, after an initial trial period with these strings and frets, the fingerboard will be planed down by about 1 mm at the nut end and the nut reworked accordingly. Graduated gut frets will then be fitted

Initial string tensions at g' pitch, A440 are 3.5 Kg. 3.3 Kg, 2.5 Kg, 2.5 Kg, 2.5 Kg. The top string tension feels too high so will be reduced accordingly.

To measure the action (i.e. the distance between the fingerboard surface and underside of a string), I used an easily made tool. Made from a piece of boxwood (but any close grained hardwood will do) cut to a slow taper. The gauge is calibrated, using a dial gauge caliper, for 2.5 to 5 mm gaps in 0.5 mm increments.
In operation, the string is lightly sounded and the gauge slid underneath until contact is made with the string (a buzz and then silence).
The action at the 7th fret position (at the neck joint) measures just over 3.5 mm on the treble side and about 3.75 mm on the bass side. This will be reduced to 2.75 mm on the treble and 3.5 mm on the bass.

So far so good - here are some images of the completed oud.

...... and a few detail images.

Dear John,
This is an amazing process and instrument! Bravo
Next step is a replica of a lute?

Take care
Yaron Naor

Hi John,

Your meticulous attention to detail is mindblowing; and your documentation of each step in the process is to be commended (duh ). Thanks for all your hard work, its very much appreciated.

Regards,

PaulO

Quote: Originally posted by jdowning

Thanks Yaron. Actually this is my first oud - my historical interest in the oud naturally developing from my primary interest in the lute. I started making replica lutes (and other plucked instruments of the 16th and 17th C) way back in the 70's. Still more research projects to go however - time and life permitting! (e.g see 'Old Project - New Lute' , posted 11-5-2007, page 8 on the Oud Projects Forum - sadly, still awaiting completion!)

I'm rather impressed by what you have achieved John, and all the research involved in this project.
Really wonderfull! As others I'm a little sad that this journey is over - it's time for the next one

Dan

Thanks DaveH.
Of course, I have played a lot of notes with the preliminary set of strings and did make some brief audio recordings (before removing the strings and frets), in order to assess and record the oud response etc using the spectrum analysis feature of "Audacity" software.

The strings used are those that I had to hand (i.e. less than ideal) - plain nylon for the first two courses, Pyramid 1010Al for the third, Pyramid 906 for fourth and Pyramid 1010 for the fifth. The plain nylon is stock that I have from the 'good old days' when the material could be purchased in 50 metre spools so may be a bit different from the plain nylon strings currently sold by Pyramid. I used 1010Al for the third course only because I did not have any suitable plain nylon (or PVF). This selection of strings was for g' tuning of the first course giving a tension of about 3.5 Kg with 2.5 Kg for the lower strings. However, at a pitch of f' sharp - that seems to suit the oud better - this tension drops to 2.8 Kg and 2.3 Kg for the lower three courses. So, not the preferred selection of strings but good enough for a preliminary assessment of the oud. I expect to hear significant improvements with the new strings that will have PVF for the top three courses at a slightly higher tension.

So, with all this in mind, here - for what it is worth - is a brief audio clip (a little doodle up and down the fingerboard) just to give some idea of the potential, as well as an image of the corresponding spectrum analysis (don't ask me what it means - it is only for file and future reference at this point in time). The strings are being plucked with fingertips - lute style.
The 'raw', unenhanced recording was made with a Zoom H2 digital recorder using the front stereo microphone pair placed about a metre from the oud. The room is pretty well 'dead' acoustically being lined with books wall to wall. The recording is compressed to MP3 format so there are consequent audio losses in the file.


Attachment: Old Oud Clip 4.mp3 (299kB)
This file has been downloaded 521 times

With temporary strings and frets removed the fingerboard has been planed down by 1 mm at the nut end and re-levelled. The nut has been reworked to bring the strings to a clearance of
1 mm above the fingerboard surface. A preliminary check of the action measures just under 3 mm at the 7th fret position (the neck joint) treble side and 3.5 mm at the bass side. These measurements may increase a little under full string tension but the action will now be left as it is.

Removal of the temporary strings reveals staining of the sound board and fingerboard surfaces below the strings - not unexpected. This is primarily due to the third course aluminium wire over spun strings (Pyramid 1010Al). These are good strings but the metal wire - being soft and readily oxidised - leaves a dark oxide deposit on the fingers, after a few minutes of playing, that is then transferred to the playing surfaces. The marks on the sound board were easily removed with a soft pencil erasure and the affected areas reworked lightly with a curved scraper blade (I avoid use of abrasive papers wherever possible - no matter how fine - when finishing wood surfaces).
The marks on the finger board were removed during the planing and levelling process.

To minimise future staining (due to oils from the fingertips), the sound board will be finished with a very thin wax coating and the finger board with a penetrating light oil sealer.

Old lute soundboards surviving from the 16th and 17th C show no signs of local staining of this kind - just an even oxidation of the wood surface. The sound boards must have been coated with an invisible protective coating of some sort - as an unfinished sound board would show localised staining over time - but what was it?
No ouds survive prior to the late 18th C but I would expect that the early oud sound boards were finished in much the same way as lutes.

Several possibilities for lute sound board finish have been proposed - but without scientific verification - including thin glue size, egg white, whole egg mixed with linseed oil, and wax.
An old Egyptian oud that I own has a sound board (and bowl) that seems to have been protected with a wax finish (the finish can be removed with solvents like 'GooGone'). Some modern lute makers also use a wax soundboard finish.

I have used very dilute shellac as a sound board sealer with some success in the past. However, as this is an experimental project, I am going to try a thin wax finish on the oud.
For this, I may use either 'Minwax' Finishing Wax or 'Lee Valley' Conservator's Wax - yet to be decided.

For the fingerboard, I shall seal the surface with a dilute, penetrating, oil based sealer. There are a number of modern alternatives possible - for example, diluted 'Danish Oil', diluted 'Minwax' Antique oil finish (the stuff used to finish the oud bowl) or diluted Tung oil. As I have a partly used can of the stuff, I am going to try 'Lee Valley' brand polymerized Tung Oil Sealer - specifically formulated as a wood grain sealer but just a dilute solution of polymerized Tung Oil.
The oil penetrates and seals but is otherwise invisible and does not leave a varnish like coating.

The fingerboard has been given a first coat of Tung oil sealer. This is applied in the same way as any rubbing oil. Applied to the finger board with a rag until the whole surface was wet, allowed to dry for about 5 minutes (until dry patches were in evidence) and then all surplus oil wiped off with a clean rag. This will be left to dry for 24 hours before a second and final coat is applied. The objective is only to penetrate and seal the wood - not to varnish the finger board. The box wood is hard and close grained to start with so two coats should be sufficient.

The sound board has been waxed with the 'Minwax' finish as the 'Conservator's Wax' seemed to be too soft for the purpose. To ensure the absolute minimum wax coating - just sufficient to protect the wood surface against staining - a small amount of the wax was placed in a piece of fine cotton fabric that was then used as a 'rubber' to apply the wax in a very controlled fashion. The first application removed quite a bit of 'dirt' (mostly fine ebony particles) and revealed a small glue spot that was then removed with a scraper blade.
A second application was then made with a clean 'rubber' and the wax buffed to a slight sheen when dried - invisible but sufficient to accentuate the fine 'cross-silk' figuring of the quarter-sawn Spruce wood grain .
The shellac coated rosettes were not waxed.

The additional fret gut and new strings from Pyramid should arrive any day now so it is time to calculate fret spacings and prepare fretting templates.

The oud will be a 'test bed' to evaluate a variety of possible fret placements (an advantage of 'tied on', adjustable gut frets).

A starting point for fretting a lute is to use equal temperament fret spacing. In this system an octave is divided into 12 equal intervals (semitones), each (for convenience) being subdivided into 100 units or cents.

In practice, theoretical fret placement (in whatever historical temperament system is used) may be affected by practical considerations such as string tension, fret diameter, string action, string diameter, string elasticity, string material, string construction, actual vibrating string length, musical mode or key etc. At the end of the day it is the player that will make the fine adjustments necessary to suite his or her judgment of the best overall 'in tune' fret placement.

Attached, for information, is the calculated fret placement for the project oud in equal temperament. The vibrating string length used in the calculations is slightly less than the distance between the front edge of the nut and front edge of the bridge because of the way each string is tied at the bridge.

This will now be compared with alternative historical (theoretical) fret placements for both lute and oud.

The gut frets and new strings have arrived from Germany so a layout of some of the possible historical fret positions (oud and lute) has been prepared - drawn full size - so that the fret positions may be readily compared and transferred to the fingerboard of the project oud.

On the left side of the Fret Position diagram attached is the fingerboard, drawn full size, with the frets located according to 12 tone Equal Temperament (or more correctly E.D.O. - Equally Divided Octave). In this system the divisions of the octave are all of equal size. So that comparisons may readily be made between various tunings and temperaments, the E.T. divisions are each divided into 100 parts or 'cents'.

Six other fretting arrangements have been represented on the diagram, calculated according to:
1) Pythagorean or Just tuning.
2) Al-Farabi's 10 fret system for oud(10th C)
3) Safi Al-Din's system for oud (13th C)
4) Hans Gerle system for lute (1532)
5) Juan Bermudo's system for vihuela (1555)
6) John Dowland's system for lute (1610)

The vibrating string length assumed for calculation purposes is 559 mm i.e. 1 mm shorter than the nut to bridge distance of 560 mm to allow for the slight shortening of the string at the bridge.

Note:
- The length of the fingerboard is 1/3 of the string length so that the fingerboard will accommodate 7 frets in the lute systems - the 7th fret being located at the neck joint.
- Al-Farabi proposed both a 10 fret and 12 fret system i.e. 10 or 12 frets up to the 5th fret in the lute system. The same applies to Safi Al-Din's system. If extended to the full length of the fingerboard, the Al-Farabi 10 fret system would consist of a total of 14 frets and the Safi Al-Din system 10 frets. It should be noted that not all of these frets would have been used by the oudists - the frets employed being the choice of the player to give the best intonation (to his or her ear) dependent upon the mode of a composition. However, both theorists stopped their fretting at the 5th fret lute system equivalent (little finger fret). This suggests that the oudists of the time did not move to higher fret positions.

More to come!

The "Varietie of Lute Lessons" was published in 1610 by Robert Dowland, son of the renowned lutenist /composer John Dowland. Included in the text are instructions for stringing and fretting a lute of the late 16th C.
To determine fret locations, Dowland first marked the positions on a strip of wood using dividers before transferring these dimensions to the lute fingerboard. This is likely how the fret locations on early uds were also determined and marked out in the halcyon days before the invention of logarithms, 'cents' and all of that "good stuff". So - for example - the fifth fret (in Pythagorean temperament) is found by dividing the total string length into four parts, the distance from the nut to the 5th fret being equal to one part (i.e. a quarter of the string length)
It should be noted that Dowland, in practice, may not have used the fretting scheme that he proposed - his lute being tuned closer to 12 TET. However, no doubt his intention was to give reasonably close fret locations as a starting point that would then be 'fine tuned' by the player 'by ear'.

Always looking for possible historical links between the Medieval oud and Renaissance lute one comment by Dowland might be of some relevance. He states:
" Now these frets of late yeeres were but seven in number, as witnesseth Hans Gerle lutenist, citizen and lute-maker of Nurenburge, (for so he stileth himself in his booke of tableture, printed in 1533) ....... Yet presently after there was added an eight fret: for myselfe was borne but thirty yeeres after Hans Gerles booke was printed, and all the lutes which I can remember used eight frets ....."
Dowland is referring to Gerle's instruction book for lute and viol of "Musica Teutch" of 1533. Checking my copy of the book, a lute is shown and described as having 8 frets (although the viols are depicted with 7 frets). Nevertheless, Dowland was likely correct in his general observation that lutes at the beginning of the 16th C were generally fitted with 7 frets.
This is confirmed in an earlier book "Musica Getutscht" by Sebastian Virdung (Basel 1511). The author discusses three types of lute - the 5 course lute (the older form of the instrument), the 6 course lute (the most common) as well as a 7 course lute (rare).
He includes a diagram of the fingerboard of the 6 course lute marked out in German tablature up to the 7th fret position. Virdung states that a lutenist can go beyond the 7th fret, but there are no fixed rules for notating it.
Virdung tells us that the tablature system was invented by Conrad Paumann, renowned organist and lute player of the 15th C.
German tablature was originally designed for a lute of the
15th C i.e. an instrument with 5 courses and 5 frets. Like the tablature for the Medieval oud each tone on the fingerboard is represented by a unique symbol. By the early 16th C, German lute tablature was adapted to accommodate a lute with 6 courses and 7 frets to the neck joint by adding new symbols to represent the fret positions 6 and 7 as well as new symbols to represent the additional 6th course. This modification can be seen in Virdung's finger board engraving (image attached - edited to show the extent of the original lute tablature).The symbols for positions higher than fret 5 are a repeat of those used for the first 5 frets only doubled 'ee' for 'e' etc. Likewise for the 6th course the symbols for the 5th course are repeated as 'upper case' figures 'A' for 'a' etc.

Just for interest here is an image of Virdung's lute player. Note the right hand position that is clearly plucking the strings with 'thumb out' technique rather than 'thumb under' generally accepted today as the style for the early 16th C.

Going back to the oud engraving. This is shown with only 7 frets on the fingerboard the last fret being the fourth or little finger fret positioned part way up the fingerboard (equivalent to the 5th fret of a lute). This is, however, consistent not only with the tuning system of Ibn Sina but also Safi Al-Din as well as the 10 fret system of Al-Farabi - the latter system providing multiple choices for the oud player (90 or 99 cents anterior, 145 or 168 cents Persian/Zalzalian anteriors, and 294 or 303 cents second finger.

The attached image of the fingerboard of the oud engraving shows the fret layout for the Ibn Sina system together with the equivalent lute fret positions. The oud tablature overlay is that of Safi Al-Din (with Roman Letters substituted in place of the original Arabic symbols - see "Safi al-Din al-Urmawi and the Theory of Music " by Dr Fazli Arslan).

The fret spacings on the oud engraving are not to scale but - surprisingly - the relative position of the fourth finger fret is quite accurately depicted.

The fretting of the oud has been completed today.

The frets have been positioned according to Ibn Sina's alternative scale including frets for the 'quarter tone' Zalzalian Anterior fret (at 139 cents) and the Zalzalian second finger fret (at 343 cents). Above the 5th (lute) fret position a sixth and seventh fret has been included as a provision to test the late 15th C 5 course lute configuration.

The first step is to transfer the calculated fret positions onto a strip of wood - the 'fret ruler' let's call it - and then use the ruler to mark the fret positions on the fingerboard in pencil. This is the procedure described by Dowland in the late 16th C.

Dowland also recommends use of graduated frets in order to keep string action as low as possible. He recommends that the first two frets (nearest the nut) should be the same diameter as the Countertenor strings of his lute, frets 3 and 4 - the diameter of the great Meane strings, frets 5 and 6 - the same diameter of the small Meane strings and the remainder of the frets the same diameter as the Treble string. Translating this into gut fret diameters, my conclusion is that frets 1 and 2 were about 0.9 mm, frets 3 and 4 - 0.7 mm, frets 5 and 6 - 0.6 mm and the remainder- 0.5 mm.
Gerle makes similar recommendations about graduated frets (not certain though as I do not have an English translation of his book - only a copy of the original on microfilm). Do the early Arab/Persian texts make similar recommendations? I do not know. However, graduated frets will be used for this project - in 0.5 mm steps from 0.5 mm to 0.9 mm.

I am using fret gut from 'Pyramid' - a good quality product available in 0.5 mm graduations.

The simplest and most economical way to tie a single (rather than double) gut fret is to use a basic 'half hitch' knot pulled as tightly as possible with the free ends of the gut cut short and melted back to the knot with a fine tipped soldering iron( to prevent the knot from slipping). The use of pliers to pull each knot as tightly as possible is necessary.
Each fret (starting at the neck joint and working towards the nut) is tied about two positions higher and then carefully slid into place using the taper of the neck to provide extra tension to each fret. The edges of the fingerboard must be smooth - to prevent a fret catching in any slight crack or crevice when being slid into position - and slightly rounded so that each fret is held tightly to the fingerboard surface under tension. A fingerboard with tiled edge is a disadvantage in this respect.
The fret knots are all arranged along the bass side of the fingerboard.
The first fret (next to the nut) is the most difficult to fit as it is the largest in diameter (0.9 mm) and stiff and as there is less space available in which to slide the fret into position in order to correctly tension it.

Checking the string action at the neck joint - after adjustments made by shaving down the fingerboard at the nut end - is
2.75 mm clearance between the underside of the Hadd (treble) string and fingerboard surface and 3.5 mm under the Bamm (bass) string.

A good question Sazi.

I have never heard of fret gut being applied wet in order to take advantage of the gut shrinking when it dries. Indeed I am not sure (like you) if gut strings do shrink in length from a wet to dry state. They certainly will absorb moisture, soften and swell in diameter but does this increase in diameter go with a reduction in length and conversely, does the shrinkage in diameter when dried go with an increase in length? I don't know the answer so - in the interests of of luthier science I am today running a simple test.

A piece of fret gut measuring 0.025 inches (0.63 mm) in diameter has been immersed in water for 5 hours. The wet gut, now fully saturated with water, is soft, pliable and quite slippery. It measures about 0.035 inches (0.89 mm) in diameter.
The length of gut has been tied between two nails under tension. The wet gut is so slippery that self tightening 'reef' knots had to be used that would hold fast under the tension.
Under tension the diameter of the gut measured 0.030 inches (0.76 mm) and the distance between the nails measured 150 mm. The gut under test has now been set aside in a warm kitchen at about 58% Relative Humidity to dry overnight. So we will see what tomorrow brings!

My guess is that the gut will reduce in diameter and expand in length upon drying. This would result in loose frets if the gut is applied when wet - but we will see.

Gut strings are made by twisting the gut when wet. Each string is then tied between two pegs on a frame and allowed to slowly dry. As the string dries the diameter reduces and the string increases in length - slackening on the frame. For several days, additional twists are applied to the string - to take up the slack - until a stable situation is reached after which time the string is allowed to dry and 'season' for a week or two before being finished.

When fret gut is tied dry, there is sufficient taper in the neck of the oud or lute to tension a fret adequately when slid into place. Wet gut would be difficult to tie, might compromise the final precision of the fret diameter after drying and otherwise may offer no other advantage?

The piece of fret gut under test has been allowed to dry for 24 hours - under tension - at a relative humidity of around 59%.
The diameter when dry has returned to its original diameter of 0.025 inch (0.63 mm). The gut still retains some tension, stretched between the two nails - but the tension seems to be less than when the wet gut was first tied to the nails (although this is a subjective judgment). However, the dried gut now feels uneven to the touch. The attached macro images show why. Fret gut is supplied smooth, uniform and cylindrical but when the gut absorbs water it softens, swells and increases in diameter. As it swells, the twisted gut filaments also start to unwind so that the original uniformity of the gut string is lost upon drying.

So, it would seem that wetting the fret gut before application is not to be recommended because:
1) the original uniformity of the gut is compromised.
2) the gut is too slippery when wet to tie a slip knot necessary for initially securing and tensioning the fret on the neck.
3) wet gut does not appear to shrink longitudinally upon drying. If anything it loses tension.

Oo-er, that's quite a difference in the smoothness and diameter there, obviously unusable, and I hadn't thought of the slipperiness either, though I should have.

Thanks for taking the time to experiment. Myth Busted!

Before declaring 'Myth busted' I am going to try one more test with wet gut to see if it might work with double frets.
Gut double frets were used historically on lutes - they can be tied a bit tighter than single frets and may wear better. A disadvantage - particularly with larger diameter frets - is that it is sometimes difficult to avoid a small gap between the frets when under tension (caused by the frets being compressed slightly as they are bent over the edge of the fingerboard).
With low string tension and low action this can result in buzzing of the string - a fault that may be avoided if a single gut fret is used.

The next test will be to wet the gut until soft and pliable and then tie a double fret. As with a single fret, a special slip knot is used to allow the fret to be tightened onto the neck and fingerboard. It will be interesting to see if the softened gut will mold itself better at the edge of the fingerboard to eliminate any gap between the frets. It is unlikely that the knot will hold under much tension (due to the slippery nature of wet gut) but if then allowed to dry, it may then be possible to apply tension in the usual way using the taper of the neck and by sliding the fret down the fingerboard.

The piece of 'fret gut' intended for test 2 (image previously posted as dry fret gut) is not gut at all but, on closer inspection, appears to be rectified nylon. The 'acid test' is that after soaking in water for 24 hours it shows no signs of absorbing water and swelling. This must have been an odd piece of instrument string that found its way into my stock of fret gut.

For information and comparison, the attached image is a macro of some high quality fret gut (Pyramid) which - like the nylon string - is uniform and smooth. The true gut when dry can be identified, under magnification, by twisted fibres that can be seen in a few places.

I have another short piece of fret gut to hand so will use this instead for the second test.

After soaking in water overnight, the fret gut has been prepared for test 2. From an original diameter dry of 0.035 inches (0.89 mm) the gut has swollen to 0.048 inches (1.22 mm) in diameter.

The softened gut was then tied to the neck of one of my spare vihuelas as a double fret - the knot configuration as seen in the attached image. It was possible to pull the knot tight to the neck without slipping but the gut was too soft and flexible to then push the wet fret down the neckin order to increase fret tension (note also that the fret fibres have unwound significantly). Due to the softness of the gut it can be seen from the images that the gap between the fret windings is even more pronounced than if the fret gut had been tied dry. After allowing the gut to dry, this discrepancy remained and could not be improved by sliding the fret down the neck to increase tension. In fact, due to shrinkage of the gut diameter upon drying, the fret was found to be fairly loose on the neck.
So - on the basis of these tests - it can be confirmed that gut frets should be tied dry and should not be first softened by soaking in water.

For information, two varieties of meantone tuning - 1/4 comma and 1/6 comma - have been plotted for comparison with the earliest surviving oud fretting arrangement given by Al-Kindi
(9th C). Note that for the first two (anterior 'mujannab')) frets on Al-Kindi's scheme of things for a 5 course oud (90 cents and 114 cents Pythagorean) - fret #1 is only used for the (bass) Bamm, Mathlath and Mathna strings and fret #2 only for the treble Zir and Hadd strings. (I have added the "missing" 6th and 7th lute equivalent frets for comparison).
The equal tempered lute fretting 'splits the difference' between the two frets for the lute fret #1 position.

According to Mark Lindley a variety of meantone fretting may have been used for the early 16th C lute repertoire - particularly in Germany and Spain. Pythagorean fretting for the lute may have been used prior to the 16th C but there is no surviving music for lute from this period to verify this speculation. Otherwise - for the solo lute repertoire of the 16th/17th C at least - equal temperament fretting (more or less) prevailed.

Quote: Originally posted by jdowning

(BTW, you reference my 'old oud' fingerboard chart (giving the Ibn Sina fretting) but I take it that you meant to link to the 'fret position' (#1) previously posted?)

I have not had time this year to make a full set of silk strings according to the directions given in the Kanz al-Tuhaf. However, in order to make some progress for this project, I do have some silk lute strings that might be used for initial tests - made for me some years ago by Alexander Rakov of New York. These are strings that have been previously used for testing but which should still have some useful life left in them. Alexander's strings are beautifully crafted - made in the fashion of the ancient Chinese strings i.e. from raw silk (with added linear mass) and with the larger diameter strings over-spun with silk windings on a twisted silk core (for extra flexibility).

I need to check to see what diameter strings I have to work with but the attached calculation sheet gives the required diameters according to the Mersenne-Taylor Law for thin vibrating strings. These calculated diameters are just a starting point and so may vary according to actual performance of the strings. For example, the linear density of the strings is likely to be somewhat greater than the 1.3 gm/cc assumed in the calculations or string tensions or overall pitch level of the strings may be varied - all helping to reduce string diameters for better performance - particularly of the Mathlath and Bamm strings.

So, let's see how it goes from here.

At this point it is time it may be informative to perform a "reality check" using the string making data provided by the 14th C. Kanz al-Tuhaf that determines string diameter according to the number of silk threads used to make each string - i.e. Hadd 16 threads, Zir 24 threads, Mathna 32 threads, Mathlath 48 threads and Bamm 64 threads.
It can be seen from the attached calculation that the diameters given by the Tanz al-Tuhaf - starting with a Hadd string of 0.44 mm diameter - are significantly less than those given by the Mersenne - Taylor Law.
This implies that the data in the Kanz al-Tuhaf may be incomplete - that there is a missing step. This possible discrepancy will be discussed in more detail later in the 'Silk Strings - Making Sense of the Historical Data' topic on this forum.
In the meantime - referring to the Mersenne - Taylor formula - it can be seen that the discrepancy in diameters might (in part or in total) be accounted for by:
1) reducing string tension as string diameter is increased (a standard procedure with modern lute strings).
2) adding linear mass to a string by soaking it in heavy metal salts ('silk weighting' as it is known in the silk industry).
3) wrapping a silk string core with a silk string or ribbon (equivalent to a modern over-spun string) to provide additional linear mass with minimal effect on lateral flexibility of a string. This is how the ancient Chinese made their larger diameter silk strings. The early Chinese texts only gave the number of threads required to make the silk core - not the wrappings. So, for example and perhaps likewise, the Bamm string in the Kanz al-Tuhaf may have had a core of 64 threads to which may have been added an unspecified mass of additional silk as a wrapping - and hence the diameter differences.
More to follow!

The oud is now fully strung with silk strings and ready for testing.

These are experimental silk strings made some years ago by Alexander Rakov of New York for my 60 cm string length lute so were not designed for this oud. Nevertheless, I was able to put together a set of strings that seem to work quite well.

The Hadd string is 0.43 mm diameter, Zir 0.52 mm diameter, Mathna 0.76 mm diameter. These are simply twisted strings made from raw silk and 'cooked' in the ancient Chinese fashion (ie with the Sericin gum intact).

For the Mathlath and Bamm strings I have used silk cored wrapped strings. The Mathlath string was originally made for the lute 5th string and the Bamm for the lute 6th string but have been 'moved up a step' on the oud (to 4th and 5th courses) to achieve a higher string tension approximately equal to that of the top three courses. The Mathlath string measures 1.08 mm outside diameter with a silk core measuring 0.54 mm diameter. The Bamm string measures
1.4 mm outside diameter with a core diameter of 0.60 mm.
None of the strings have been 'weighted' hence the relatively large diameter of the basses (compared to modern wire wound strings).

The strings are tuned a fourth apart with the top string at maximum workable pitch of f sharp (370 Hz @A440). (I did try taking the top string to G but one of the Hadd strings failed after about a week - so that pitch is judged to be the upper limit).
Tuning the top string to maximum pitch (avoiding frequent breakage) allows the thicker bass strings to respond better at higher tension.
The string diameters available do not give exactly equal tension but are close enough ('by feel') for initial trials. (Calculated tensions are Hadd 3.5 Kg, Zir 2.7 Kg and Mathna 3.3 Kg).

The wrapped strings have been beautifully constructed by Alexander - looking like a 'string of pearls' according to the ancient Chinese string makers.

Sound clip to follow.

The oud has been fitted with graduated diameter tied on gut frets (reducing in diameter from nut to neck joint) to keep the 'action' as low as possible. For initial trials with silk strings the frets (total 7 to the neck joint) have been spaced according to Western 12 tone equal temperament - familiar territory for now - after which the various historical Pythagorean/Arabic/Persian systems (with intermediate fret positions - or 'quarter tones') will then be tested for comparison. The five double courses are tuned a fourth apart.

I have decided to use the more historically appropriate plectrum (mizrab) for the trials - rather than finger style plucking. I have no previous experience of using a plectrum so this will be a useful learning experience.

To minimise wear and tear on the silk strings a mizrab has been cut from a strip of thick, hard cow hide leather - softer and gentler than a feather quill (risha) or one made from bone, horn or plastic. I figure that the leather is likely similar to tree bark that was once used in ancient times.

So here is the first sound clip of the oud strung completely in silk - plucked inexpertly with a leather mizrab - just to demonstrate the range from open bass (Bamm string) to the top string (Hadd) as far as the neck joint (7th fret).
The sound recording has been made in an acoustically absorbent room (lined wall to wall, floor to ceiling with books). The digital recorder (Zoom H2) was positioned about a metre from the oud. Audacity software was used to select the sound clip and convert to MP3 - otherwise the audio recording is 'raw' - as is - and unedited.

For such a small instrument (56 cm string length), the projected sound volume and bass response in particular is (for me) encouraging. Also the trebles 'sing' with a bright 'silvery' tone that may be a characteristic of silk strings.
Checking earlier postings - when the topic of Helmholz resonances was discussed (page 10) - it was estimated that the Helmholz resonance of the oud body might be optimal at a frequency of around 169 Hertz. This falls close enough to the 4th string (Mathlath) frequency of 156 Hertz to add support to the Helmholz resonance theory (at least as applied to this oud).

The strings were plucked at a location mid point between the two sound holes which seems to give optimum results re: dynamic range of sound volume. So the wide spacing of the two relatively large diameter sound holes may make historical sense - to avoid damage to the sound holes from the plectrum perhaps?







Attachment: Old Oud Silk String AudioTest 1.mp3 (314kB)
This file has been downloaded 552 times

The broken pieces of the Hadd strings have been subject to testing to determine the breaking load and stress - useful information in order to establish optimum string design.

The broken string pieces were subject to a straight pull using a simple spring balance to apply a loading. The spring balance is not a very precise piece of equipment so was calibrated (approximately) using standard weights from a commercial grocer's scale.
After several tests, the average breaking load was established at about 5 Kg (11 lbs).
The string diameter is 0.43 mm which gives a calculated breaking stress of about 35 Kg/sq. mm. (49,000 p.s.i.)
Interestingly this is about the same as the breaking stress for gut trebles of the 16th and 17th C (average value
34 Kg/ sq. mm. - source Mimmo Peruffo, 'The Lute in its Historical Reality').

Using the Mersenne-Taylor Law, if the strings were raised to a pitch of a' sharp @ A440 (frequency 466 Hz) the strings would be above breaking load (5.24 Kg). Practically (historical experience), the pitch of the trebles should be set at about
2 -3 semitones below this theoretical value (for gut strings at least) in order to avoid frequent string breakage. This practical rule of thumb would bring the maximum working pitch of the Hadd strings to a bit less than g' @ A440 (392 Hz).
So f' sharp @ A440 (370 Hz) should also be about the optimum pitch for the silk Hadd strings.

The degree of twist of the silk filaments in string construction affects the ultimate strength of a string - the smaller the angle of twist (low twist) the greater the breaking load and ultimate breaking stress. The attached macro image of the Hadd strings indicates a twist angle of about 30 degrees. Further reduction of this degree of twist (consistent with producing a uniform cylindrical string) would produce a string of higher tensile strength.
String durability will also depend upon the abrasion resistance of a string - as individual filament are broken by the action of a mizrab or risha - or by localised wear on a fret - then the strength of a string is reduced accordingly.