Silk Oud Strings - Making Sense of the Historical Data

H!

Awesome Thread!

I wonder How Silk strings feel to your fingers.

Is This Only Handmade or can be Bought in a Store?

Thanks

Philip

There is no question that oud strings - according to the earliest records dating from the 9th to 14th C - were made either from silk or gut (sheep's intestines).
(As the European lute was a development of the oud it would seem likely that lutes were also strung with the same materials. There are some who would deny that silk strings were ever used on lutes but that is another story not relevant this discussion!).

In order to better understand the possibilities (and difficulties) in trying to replicate historical oud strings from silk, it will be necessary to start from the beginning.

Silk filament is produced by a variety of moth caterpillars to spin their cocoons - although only a few are of commercial interest. The species of moth of primary interest for string making is the Bombyx mori - domesticated by the Chinese exclusively for silk textile production many thousands of years ago.

The silkworm (caterpillar) produces the cocoon filament from two spinnerets in its head. The cocoon raw silk filament or Bave is composed of two strands (Brin) of protein (Fibroin) cemented together with a natural gum (Sericin).
Sericin and Fibroin are of about the same Specific Gravity, Sericin accounts for about 25 - 30% of the total weight of a raw silk cocoon filament. Sericin is soluble in hot water, Fibroin is not.
Silk fibroin is elastic and proportionally stronger than steel.

The cross section of the Bombyx mori Brin is approximately an equilateral triangular and that of the Bave, approximately elliptical - however, there are dimensional and geometrical variations - silk being an organic material.

The attached images show an idealised, model Brin section - measuring, typically (for modern silk filament), 10 microns (0.01 mm) in 'diameter' - and an associated model of the Bave measuring about 25 microns (0.025 mm) in 'diameter'.
The area in red colour represents the Sericin gum.

A micron is a millionth of a metre so we are dealing with microscopic dimensions finer than a human hair.

To illustrate the variability of cocoon silk filament, the attached image is a plot of five selected cocoon samples showing the variation of weight (in grams per 100 metres) for each 100 metres of reeled cocoon filament and the 'diameter' of the Bave measured at the midpoint of the total length of reeled filament in each case. The weight per 100 metres is an indication of the proportional 'diameter' variation of the cocoon filament.
The source of this data is from "The Royal Jubilee Exhibition, Manchester, 1887" with cocoon samples from Japan, France, and the, then, British Colonies of Victoria (Australia), Ceylon (Sri Lanka), Cyprus.

The strains of Bombyx mori that produced these samples may now be extinct - just to add to our difficulties! Note that the measured 'diameters' of Brin exceed (significantly) those of modern raw silk (with a mean 'diameter' of 23 microns or 0.023mm) reported by modern researchers.

This is going to be a bit more complicated than at first anticipated!

Regardless of the cross section 'diameter' of a single raw silk cocoon filament - be it 23 microns or 50 microns or more - a single filament is too fine for practical use.

The traditional procedure then is to combine several cocoon filaments to produce a thread of greater diameter. This is achieved by reeling together the filaments from the required number of cocoons needed to produce a desired thread diameter.

Hand reeling is a highly skilled operation.
The cocoons are first soaked in hot water to soften the Sericin (some Sericin is also removed in the process - dissolved in the hot water) allowing the filament to be unwound in a continuous length. After locating the filament end of each cocoon, the filaments are together wound onto a reel after passing over a pulley system that compresses the filaments together. The Sericin gum quickly re- hardens to cement the filaments into a uniform thread.
Raw silk destined for the textile industry is reeled into threads many kilometres in length. Due to the limited length of a cocoon filament and variations in filament 'diameter' along its length, new cocoon filaments must be continuously added as reeling commences - the skill of the operator is to judge when additional filaments must be added in order to maintain the required thread diameter (within a tolerance limit specified by the textile industry).

Early Chinese texts giving details about making strings for the ancient Chinese zither state that a thread was made from 12 cocoon filaments. So let's assume - as a starting point - that this was also the filament count for the threads used to make the early silk oud strings.
Using a model of a thread composed of 12 cocoon filaments of an idealised geometry (or 24 equal 'diameter' Brin, of equilateral cross section, each measuring 10 microns) to illustrate the possibilities.
The attached image shows the most compact assembly of the Brin which measures about 47 microns or 0.047 mm in diameter. The spaces in between the Brin elements are filled with the melted and re-hardened Sericin gum.

For information, a good photo of a reeled raw silk thread - taken with a scanning electron microscope - can be found in the APPENDIX, image #4 of the "Silk Reeling and Testing Manual" at

http://www.fao.org/docrep/x2099e/x2099e00.HTM

With the cocoon filaments combined together to make a thread, the individual threads can now be twisted together to make a string.
According to the author of the 14th C Kantz al-Tuhaf, a silk 'Hadd' string was made from 16 threads. For a 56 cm string length we know that the string diameter, in either gut or silk, should be about 0.4 mm with the string at maximum tension.

Again, using a scale model as an illustration, the attached image shows a 'Hadd' string cross section made from up from 12 filament threads with Brin 'diameters' of 10 microns. Here it can be seen that the most compact arrangement of the 16 threads gives a string diameter of only 210 microns or 0.210 mm.
To create a uniform cylindrical string the threads must be tightly twisted together - a process that will compact the filaments together further reducing the string diameter (by an unknown amount at present).
An increasing degree of twist increases the flexibility and elasticity of a string but with a consequential reduction in the string breaking stress.
In a simply twisted string, the outer threads are stressed more than the inner threads so this is a construction best used for the thinnest of strings. Once twisted, the elements of a string made using this construction must be locked in position to prevent unravelling. If raw silk is used, the Sericin may be used to hold everything in place by first softening the gum prior to twisting a string.

A better solution for larger diameter strings is to first bundle the threads equally into 'strands' - in this case either two strands or four strands. The strands are then first twisted individually in one direction and then combined together with twist in the reverse direction. This is the same method used for making small ropes or cords and results in a stable, uniform assembly with no tendency to unwind.
The attached images show a scale model of both 2 strand and 4 strand construction prior to twisting.

It is apparent from the scale models that a string made using a typical modern commercial raw silk cocoon filament (with a Brin of 10 microns 'diameter') cannot give the required string diameter of 0.4 mm if the historical data is followed. Scaling up, and using the maximum mean 'diameter' Brin of modern cocoon silk of around 13 microns results in a 16 thread 'Hadd' string measuring still only about 272 microns or 0.27 mm in diameter prior to twisting.

In order to make a string of 0.4 mm diameter, the Brin 'diameter' of the thread would need to be about 19 microns.
Clearly, the 14th C string makers had at their disposal threads made from larger 'diameter' cocoon filaments than those from typical commercially available modern cocoons. Is this possible? The data from the Manchester Exhibition cocoon samples indicates that this very likely was the case.

The question is are Bombyx mori cocoons, yielding this size of filament, available today? A question that, no doubt, is going to be difficult to resolve given the variability of cocoon size and quality not to mention the effect of environmental conditions resulting - like wine - in some good cocoon years and some not so good cocoon years. And how does one find cocoons of the best 'vintage' anyway?
This is not going to be easy!

The silk string maker has a choice - to use raw silk or de-gummed silk. De- gummed silk is likely easier to twist into a uniform string (less friction between the filaments), however, removal of the Sericin results in a significant density loss - a negative for a string maker. However, this weight loss may be recovered by 'weighting' of a string after construction.

Removal of the Sericin may be achieved by soaking the raw silk in a hot acidic or alkaline solution. However, as strong acidic or alkaline solutions will dissolve the silk Fibroin the solution concentration (acid or alkali), temperature, exposure time etc. is critical.

The attached image is a graph representing % Sericin weight loss of raw silk when treated with hot acidic/alkaline solutions. The pH value is a logarithmic scale of acidity or alkanility. Pure water has a pH value of about 7. For comparison, lemon juice has a pH value of about 2.2, orange juice about 3.7, modern acidic rain (!) about 5.6, baking soda solution about 8.2, soap solution about 10, washing soda (sodium carbonate) about 11 and household lye solution (Sodium Hydroxide) about 13.6.

The fibroin is not subject to damage for pH values between about 4 and 8.

No information is given in the early texts about how much wood ash must be added to the water used to remove Sericin gum from raw silk. Therefore, some brief trial were made to try to establish the alkaline effect wood ash has when added to water.

Wood ash - the residue remaining after burning wood - is a material of widely varying composition dependent upon species of tree, combustion conditions etc. The ash is comprised primarily of Calcium Carbonate (lime) 25% to 45% as well as potash (Potassium Carbonate) less than 10% and Phosphates (phosphoric acid salts) less than 1% plus trace elements. Once used as a raw material for soap and glass making - wood ash was one of the first agricultural products exported from the pioneering settlements in Eastern Canada where efficient clearance of forested land for agriculture entailed felling and burning of trees.

At this time of year we heat our home with a wood stove fuelled with white pine logs from our wood lot - so the wood ash used for these trials is softwood ash. The ash was first passed through a fine sieve to remove charcoal, clinker lumps etc. The ash was then measured by volume and by weight.
The ash was added in increments to a stainless steel pan containing 500 ml of tap water brought to boiling point after each addition of ash (do not use aluminium pans as this metal is dissolved by alkaline solutions). The incremental steps each measured (in total) 5, 10, 15, 20 and 30 grams (10 grams of ash is equivalent to about one tablespoonful in volumetric measurement). After each step in adding ash to the water, the solution was allowed to settle for a few minutes so that the ash residue had settled to the bottom of the pan before measuring pH..

The pH value of the solution was measured at each step using wide range pH test papers ranging from pH 1 to pH 14. The test papers or indicator strips are paper strips saturated in a chemical that changes colour according to the level of acidity or alkalinity of a solution. The pH level is measured by comparing the colour of a test strip with a standard colour chart. Wide ranging test papers are not a highly accurate means of measuring pH but are good enough for these trials. It should be noted that the pH scale is logarithmic so that, for example pH 9 is 10 times more alkaline than pH 8 and pH 10 is 1000 times more alkaline than pH 7 (10X10X10 = 1000)

Test results - the tap water tested at pH 7 (about where it should be). No significant change in pH was detected until the water contained 20 grams (2 tablespoonfuls) of wood ash when the indicator strips showed a pH of about 12. Water containing 30 grams (3 tablespoonfuls) of ash showed a pH of about 13. Alkaline levels of pH 12/13 can damage silk fibroin if time of immersion of the raw silk is not carefully controlled. These caustic levels present a safety hazard so plastic gloves and eye protection should be worn when working with these strongly alkaline solutions.

It has been found, by practical experience of others, that a hot (not boiling) solution of around pH 10 will de-gum raw silk in about 30 minutes without damaging the fibroin. From which it might be concluded that about a tablespoonful of softwood ash (10 grams) added to 500 ml of water, brought to simmering point, should be about the concentration required to de-gum raw silk in about 30 minutes.

Gum Arabic (or Gum Acacia) is a natural,sugary, edible, vegetable gum harvested from acacia trees in North Africa. Like all true gums it is soluble in water to make a viscous fluid or jelly-like paste and has wide ranging commercial uses. It is used, for example, as a flexible binder for water colour pigments and for making hard rubbery 'chewy' type confections like 'jelly beans', 'gumdrops', or 'chewing gum' etc.

The recorded use of Gum Arabic dates back about 5000 years to the times of the Ancient Egyptians. It is the oldest and best known of all the natural gums.

Gum Arabic has a specific gravity ranging from 1.35 to 1.49 so has a greater density than Sericin gum. Adding this gum as a binder to the de-gummed silk strings would restore any weight lost and likely be more flexible than the more brittle Sericin. Elasticity is good for instrument strings.

For this string making experiment I shall be using standard grade, unprocessed Gum Arabic that I already have in stock - see the attached image. The gum is rather attractive with its variable 'amber' like colours. Rather like raw, uncut gemstones!

Gum Arabic in its, more mundane, powdered purified state may be obtained from art supply houses.

The other component of the gum wiped onto the finished silk strings is "a little essence of Saffron". Essence of Saffron is taken to mean the 'essential oil' of the Saffron flower (Safranal) obtained by distillation of the stigmas of the Saffron Crocus flower. This is a highly concentrated volatile oil.
A dilute solution of the oil may also be prepared by simply boiling the Saffron stigmas in water.

The herb Saffron is the stigma of the Saffron Crocus flower. There are three stigmas per flower so it takes from 50,000 to 75,000 flowers to make a pound (0.45 Kg) of Saffron - all picked by hand! Current retail value of Saffron is about $300 US an ounce (28 grams) the most costly herb available on the market today (as it was in ancient times).
Cultivated in ancient Persia it was introduced to Spain by the Moors in the 8th C where it is still an important agricultural export. The name comes via Persian/Farsi (za feran) to Arabic (za faran) derived from the adjective (asfar) meaning 'yellow' (source Wikipedia).
Saffron is used today primarily in cooking due to its distinctive aroma and as a food dye - imparting a golden yellow colour to food. It also has medicinal applications.

The attached images show a variety of the beautiful Saffron Crocus with its three red coloured stigma (source Wikipedia) and the dried Saffron herb sample that I will be using for the string making experiments. The image of the dried herb measures about 3 cm across and represents about a gram in weight (a teaspoonful) costing $10! Also, I doubt if this example is top grade material.

Why do the instructions in the Kanz al-Tuhaf require addition of the concentrate 'essence of Saffron" to the Gum Arabic? The text does not say but later, when discussing the preparation of gut strings for the oud, goes on to describe the finishing process where "The (gut) strings are stained with Saffron or whitewash this being rubbed into the strings until they are dry"
So, it would seem, that the Saffron here is being used as a dye - to colour the Gum Arabic and, hence the silk strings a golden yellow?
Nevertheless, perhaps there is another purpose based simply upon the intrinsic value of the herb or , more deeply, some of its other ancient traditional applications?
Or maybe oil of Saffron has some preservative properties when added to the gum to ensure a longer string life?

For this string making experiment, Saffron will be added to hot water to dye the solution yellow. This water will then be used to make a paste of the Gum Arabic - to be rubbed into the finished silk strings.

The 14thC Kanz al-Tuhaf gives the most detailed but brief account of silk string construction for a five course oud.
The string maker of that period in time seems to have started by de-gumming commercially available threads made up from raw cocoon silk filament - threads that likely would have normally been destined for weaving into fabric.

The number of threads twisted into each size of string is given as follows:

HADD 16 threads
ZIR 24 threads
MATHNA 32 threads
MATHLATH 48 threads
BAMM 64 threads

It is not known how many cocoon filaments were in each thread used by the string makers - commercial raw silk threads might be reeled from any number of cocoons (as it is today) dependant upon the thickness of thread required.

The number of filaments per thread was important to the early Chinese string maker.Their early texts state that the threads used to make strings for the qin (Zither) should contain an optimum 12 cocoon filaments - although the size ('diameter') of the filaments used by the Chinese is not known. The 'diameter' of the raw silk cocoon filaments could well have been much greater than typical cocoon filament available today.
The previously posted idealised model of a HADD string made from 16 threads demonstrates that a thread made from only 12 de-gummed typical modern cocoon filaments cannot give the required string diameter of 0.4 mm.

The attached calculation sheet derives the relative diameters of the Kanz al-Tuhaf strings based upon an assumed diameter of 0.4 mm for the HADD string. Using the idealised model, the de-gummed thread diameter works out at around 0.10 mm (100 microns) which is about twice the diameter of a thread made from 12 modern cocoon filaments.

There are, therefore, two possible alternatives available to the modern silk string maker :

1) Starting with Bombyx mori cocoons, reel the silk filament from the required number of cocoons and use only the first 200 to 300 metres of filament for making the strings - this part being the largest diameter within a cocoon. The aim is to use the minimum number of filaments to make a de-gummed thread measuring 100 microns in diameter. This number would have to be established by trial and error.

2) Use commercially available raw silk thread - measuring 100 microns in diameter after de-gumming - as a starting point. This thread might contain considerably more smaller 'diameter' filaments than 12, and is a number that might vary from year to year dependent upon cocoon quality and yield for a particular year.

The string making experiments will be undertaken using both alternatives. The cocoons have been shipped already but some basic reeling apparatus will have to be made first - as well as apparatus for twisting the threads into strings.

The silk required for the string making trials has arrived!
I ordered 100 Bombyx mori cocoons from Treenway Silks. BC, Canada at

http://www.treenwaysilks.com

Enquiring about supply of raw silk filament, the company did not have in stock the reeled filament size that I was interested in so contacted their supplier in China to see what was commercially available. As a result - and because I was (and still am) not exactly sure how silk measured in denier will convert to microns diameter after de-gumming - their supplier kindly sent a very generous free sample of their raw silk to try. The label on the hank of silk reads 40/44 that I assume means 40/44 denier but this will have to be confirmed. Certainly enough silk to carry out meaningful comparative trials. Thank you Treenway Silks for your interest and help in making this a promising start to the investigation.

In order to handle the silk, the hank will first have to be reeled on to a bobbin otherwise it will just become a tangled mess. So the next job is to make a winding apparatus to do this work.

The raw silk appears - to my inexpert eye - to match the Kantz al-Tuhaf description that " the threads should be white, smooth, of equal gauge and well finished"

Silk thread is measured in Denier - equal to a length of 9000 metres weighing 1 gram (or 450 metres weighing 0.05 gram).

Converting silk thread in Denier measurement to diameter is complicated due to the variation in the density and 'diameter' of the silk filaments that are combined together to make a thread. For this reason silk thread measured in Denier reflects this variation with a maximum and minimum value, for example a 10-12 Denier thread has an average Denier count of 11. The finest practical thread is about Denier 6 - 8.

The attached graph of thread diameter versus Denier has been drawn from data published in "Raw Silk Properties, Classification of Raw Silk and Silk Throwing" by Warren P. Seem, 1922. (this book is available as free download from the Internet Archive website).

From this graph it can be seen that the silk sample from 'Treenway Silks' - measuring 40/44 Denier (average Denier count of 42) - has a mean diameter of about 97 microns or 0.097mm (i.e. a little less than the calculated required thread diameter of 0.10 mm for this project but close enough).
We are dealing with very small dimensions 97 microns being the equivalent of 0.0038 inch or less than 4 thou.

I now realise that there should be little difference between the diameter of a raw silk thread and a thread that has been de-gummed as the multiple seracin coated filaments of a raw silk thread are drawn tightly together during the reeling process and the sericin gum - being in a softened state - is just compressed to fill the gaps between the fibroin fibres (see the idealised models previously posted.

Just to further clarify the question of Denier versus thread diameter illustrated by taking the sample skein (or hank) of silk that will be used for this project as an example.

The skein is graded as 40/44 Denier and weighs about 22 grams (measured on a digital kitchen scale - resolution 1 gram).

The silk, therefore, varies somewhere between 40 grams and 44 grams per 9000 metres along the total length of the thread.
Diameter (from the conversion table), therefore, can vary between 93 microns and 98 microns along the total length of the thread.
The length of thread in the sample skein will, therefore measure somewhere between 22/40 X 9000 and 22/44 X 9000 or between 4500 metres and 4950 metres. More than enough for these experiments.

However, in order to use the silk thread, it must first be wound from the skein on to a bobbin or reel. In order to do this, a winding machine must be constructed. I shall be using a design published in "Silk, its production and manufacture" by Luther Hooper (the book is available as a free download from the Internet Archive website). These old texts are often useful in providing information about hand production methods and equipment.

The attached image illustrates the machine - made mostly from wood.
The skein is first mounted on a freely rotating 'swift' (A), the diameter of the 'swift' being adjustable by moving thread cross-ties on the thin 'spring loaded' spokes (2 and 3).
The bobbin (G) rotates in a groove in the frame of the machine. Its steel spindle has a small lead sleeve at one end (E) that rests, under its own weight, on a leather covered drive wheel. This provides a low friction drive for the bobbin essential to avoid breakage of the fine silk thread during winding. If the thread should tangle on the 'swift' the bobbin will slip on the drive wheel preventing the thread breaking.
In order for the thread to be wound evenly on the bobbin it passes through a wire loop (F) that moves from side to side with an oscillating motion driven by a crank connected to the main drive of the machine.

A fairly straightforward task to build but this work will have to wait until the abnormal sub zero weather conditions for this time of year improve as they should do next week (hopefully). The outside temperature this morning is minus 24 Celsius with the workshop at minus 8 C.

Quote: Originally posted by jdowning

... suspect that 'essence of saffron' also had some kind of medicinal anti-bacterial property perhaps preventing (or delaying) deterioration of the gum arabic glue used to bind the silk threads together?

Does alcohol precipitate a gum deposit if added to a solution of Gum Arabic in water?

Four small pieces of raw Gum Arabic were covered with 10 ml of water and left overnight to dissolve in a warm environment. After 18 hours, the gum had completely dissolved in the water to form a clear amber coloured, semi viscous liquid. The volume of the gum dissolved in water solution was 15 ml.
(the measuring syringe has been recycled from an ink-jet printer refill kit)

The liquid gum was applied to paper strips to test for adhesion and flexibility. On drying the paper strips had curled confirming that the gum shrinks when dry. On flexing the paper, the gum coating exhibited numerous fine cracks - so the gum is brittle.
When the gum coating was moistened with water, the cracks disappeared as the water soluble gum was reconstituted.
Paper strips glued together indicated excellent adhesive properties of the gum - the gum being stronger than the paper.
Nothing very surprising as one application of Gum Arabic is for gummed postage stamps and envelopes.

Pure Methanol (Wood Alcohol) was added to a sample of the liquid gum - little by little. Each drop of Methanol added formed a white residue on the surface of the liquid gum. On standing, the residue dissolved to form a clear liquid with similar properties of adhesion and flexibility as the original gum sample. After standing for several hours, a gelatinous skin had formed on the surface of the gum that dissolved to a liquid on stirring (surface evaporation of the methanol?)
Not sure what to make of this but there is obviously no sudden precipitation of gum when alcohol is added to the gum solution - so will observe this sample for a while longer, just to see what happens.

The authorities say that Gum Arabic is insoluble in pure alcohol so as a test a small piece of the gum has been left to soak overnight in 2 ml of Methanol.
Methanol (and Ethanol) is miscible (soluble) in water - just to complicate matters.

The small sample of Gum Arabic in water solution with Methanol added was left in an open container overnight. By this time the solution had turned to a syrupy gel consistency - most likely due to evaporation of the alcohol and water rather than the alcohol causing precipitation of the gum?
However, as a further test to confirm this, a sample of the gum in water solution will be left overnight in an open container to see what happens.

A coating of the gelled water/methanol gum was applied to a strip of paper and allowed to dry. On flexing the paper the gum layer cracked confirming that Methanol does not impart properties of flexibility to the gum. At this stage it is uncertain if flexibility of the gum as a binder for silk strings is a requirement as, apparently, in the the textile industry, Gum Arabic is used to make yarn stronger and increase its tensile strength (presumably without sacrificing flexibility of the yarn?). Good for silk strings also!

The sample of gum in Methanol has not dissolved after 24 hours confirming that Gum Arabic is insoluble in pure alcohol. However the gum is said to dissolve in an aqueous solution of ethyl alcohol (ethanol) up to 60% concentration (in water). (Possibly also in a 60% aqueous solution of Methanol?) Ethanol can also be used to extract 'essential oils' from plant material so while 'Tincture of Saffron' - being a pure alcohol solution - would not result in Gum Arabic being dissolved to a 'paste' consistency a saffron 'essence' made by mixing saffron with aqueous alcohol probably would.

So the next test will be to dissolve a sample of the gum in aqueous Ethanol. Aqueous Ethanol is readily available in small quantities from the local brewers retail outlet - I shall be using a 50ml. 'miniature' bottle of 'Smirnoff' vodka. This is a pure ethanol spirit of 40% alcohol by volume in water so should work. Any vodka left over will not be wasted!

Gum Arabic is unique among natural gums due to its extremely high solubility in water - yielding highly viscous, gel like solutions at up to 50% concentration.
This 'paste like' consistency of the gum, needed for silk string making - according to the tests so far - might most easily be achieved by simply allowing a Gum Arabic solution to stand in an open container until the volatile components (water and alcohol) have evaporated to a sufficient level to form a 'paste' convenient for rubbing into a silk string.
Of course, once the gum paste has dried out it will harden and become brittle.

The sample of gum in water solution - left overnight in an open container - by this morning had turned into a thick, sticky gel
(like the gum in Methanol sample did) due to evaporation of the water.

By this time, the gum in Methanol sample had completely hardened. This was readily dissolved back into a clear liquid gum by addition of 3 ml. of water - taking about an hour to fully dissolve. So the whole process is reversible.

A test sample of Gum Arabic in 40% Ethanol (Vodka) has been prepared and left to soak overnight. It is anticipated that the gum will dissolve in the aqueous alcohol due to the 60% water content.

After 24 hours of soaking in Vodka the Gum Arabic is dissolving only very slowly and is forming a slightly 'milky' liquid. So this does not look very promising as an avenue of investigation. I suppose that the water molecules mixed with the alcohol molecules are less free to dissolve the gum as free water in solution would have done.

Some experimentation will be necessary to determine the correct consistency of the gum/water solution to match the Kanz al-Tuhaf string makers description of "A paste of moderate consistency" - if the gum solution is too sticky and viscous it may be impossible to "rub into the strings with a piece of linen until it has penetrated into all the parts ..."?
However, Gum Arabic is an interesting material in solution. At concentrations in water of up to 40% it flows like water and other Newtonian fluids. However at concentrations over 40% it takes on pseudoplastic characteristics denoted by a decrease in viscosity with increasing shearing stress. In other words, a gum in a stiff and viscous concentration - will become more fluid when rubbed or stirred - perhaps sufficient to penetrate the innermost fibres of a string? Just another theory to test in practice!

After soaking for 24 hours at room temperature both the 40% Ethanol solution and the 99.9% Methanol have extracted dye from the Saffron.
The attached image compares the attractive transparent colours of the tinctures against a Gum Arabic solution in water.
The darker orange/gold tint of the 40% Ethanol sample is on the left the Gum Arabic sample in the centre and the paler
lemon /gold tint of the Methanol sample on the right.
At this stage, it would seem that the Ethanol based tincture would likely be more effective as a dye for the Gum Arabic than the Methanol based tincture (safer too).

A test rig for twisting the silk 'threads' into 'strands' and then from 'strands' into instrument strings is now complete and ready to run. From this point, data must be gained by trial and error so the rig design will enable me not only to gain experience on twisting and treating the strings (degree of twist, application of gum etc.) but also to undertake mechanical tests to measure stress/strain characteristics, breaking loads etc.
The design of the rig will enable construction of simply twisted as well as strings of cord or roped construction.

The rig is basically a vertical board with a geared winding mechanism at one end (an old hand drill with about 3.5:1 gear ratio) and a loaded sliding carriage at the other. At this point I do not know the loading to apply for each string diameter or the number of turns needed for a particular string construction - but I assume that the number of turns could be fairly high, hence the geared winding mechanism. String load can be varied by adding lead weights to the box on the sliding carriage. The carriage must slide freely as the silk is twisted when wet so - being hygroscopic will stretch under load by up to about 25% increase in length. The twisting operation, on the other hand, causes the string to reduce in length from its original untwisted state.
The first strings (Hadd strings) will be simply twisted to form the 'threads' into a uniform cylinder - necessary for an instrument string. The twisting will make the completed string more flexible but will reduce the string load bearing capacity. (The strongest string would be one where the 'threads' are laid in parallel without any twist at all. This was the construction used by Asiatic bow makers for their powerful reflex bows - the 'threads' being held together by wrappings at points of heavy wear and by short bindings at regular intervals along the length of a bow string).

Once data has been obtained from this rather rough and ready test rig a new, improved design may be developed for twisting the strings.

The early Chinese string makers used a somewhat simpler device for twisting their strings. Next - to review the information given by the early Chinese texts to determine the apparatus and methodology that they employed for string making - information that might have a useful application in these trials.

CORRECTION! Terminology in this posting has been revised to ensure consistency throughout this topic. So for example where applicable silk 'strand' has been changed to 'thread'.

To summarise the definitions applicable to this topic:
A silk 'thread' is made by combining several cocoon filaments together (9, 12 or more - dependant upon filament 'diameter').
A silk 'strand is made by twisting several silk 'threads' together (dependant upon the finished diameter required - 16 or more in the case of the ancient oud strings).
A silk strings of larger diameter may be two three or four 'strands' twisted together into a single cylindrical cord (or rope like) construction. The multi-strand construction of a string results in a stable combination of the silk filaments with no tendency to unwind (i.e. like a rope).

The smallest diameter strings like the 16 'thread' Hadd strings may possibly have been just a single, simply twisted strand held together by gum to prevent unwinding (i.e. when not fitted to an instrument where unwinding is prevented mechanically by a string being tied to the bridge at one end and a peg at the other). This form construction has yet to be confirmed for viability by testing. On the other hand, these strings could also have been of 2 strand cord construction so this possibility will be tested as well for comparison.

John Thompson's website on the Chinese Zither or Qin has important translations of two early books about the Qin - including sections on making silk strings for the instrument.

http://www.silkqin.com

The first book from the 15th C or earlier (Ming Dynasty) is 'Taiyin Daquanji' and the second, 'Yuguzhai Qinpu' dates from the mid 19th C. but which confirms that early string making methods were still being practiced at this late period.

The 'Taiyin Daquanji' describes an apparatus used for twisting the strings called a 'Zhuizi' (literally 'earrings'). This is a very simple tool - basically a weight, that is hung on the silk thread bundles, spun by hand to twist the threads into a strand. The silk threads 94 feet (30 metres) in length are attached to a hook on a suitably tall structure(!)
The design of the 'Zhuizi' is not absolutely clear from the translation into English so some guesswork is required to determine the exact design and mode of operation.

The device is made from two discs of fruit wood (jujube tree) - "the size being that of four coins" - fitted with a handle and - "on four sides use small bamboo pegs".
Weight is added using "small coins" - "each silk 'thread' (i.e. comprising 12 cocoon filaments) should carry the weight of four coins". "If the thickest string (for example) has 160 'threads' these will be divided into four equal bundles (i.e. to make a four 'strand' string) so each 'strand' bundle will have 40 'threads'. Each 'Zhui' will then weigh 160 'cash' (wen) and the four 'Zhui' together (i.e. one for each 'strand' bundle) will weigh 640 cash" ..... "First place the coins on the lower board. The bamboo pegs go on top.Then with the upper board press this down together".
(I have added comments in parenthesis for purposes of clarity of meaning).

The attached image is how I visualise the design of the 'Zhui' (or twirl) from this description. The device does bare some resemblance to the pendant style earrings that were popular in ancient China - hence the nickname "Zhuizi" presumably.
The top of the twirl is free to slide up or down on the four wooden pins and the central rod to allow addition or subtraction of coins - dependent upon string size required.
The central rod forms a handle underneath the base and has a hole at the top allowing attachment of a (removable) suspension hook (like the hook on an earring)

The Chinese coins used for weights ('cash' or 'wen' - more on these later) have a square hole in the centre. These are slid on the pins in equal stacks of coins distributing the weight equally.
The attached plan view shows the arrangement of coins (to scale) on the twirl base plate. Overall diameter is about 8 cm.
The coin shown (not to the same scale) is a 1 cash denomination bronze coin from the Ming Dynasty - diameter about 24 mm.

In order to determine the total weight of a 'Zhui' according to the number of 'threads' in a 'strand' to be twisted we need to know the weight of a Chinese 'cash' coin at the time 'Taiyin Daquanji' was written.
The English term 'cash' refers to coinage that was legal tender in China from the second century B.C. until about the mid
20th C.
The coin is typically round with a square hole in the centre (see the previous posting) and cast from bronze (other metals and alloys were also sometimes used but are not considered here).
As the coin had small monetary value they were issued in strings (like a necklace) with typically 1,000 coins to a string - hence the need for a hole in the middle of the coin.
The coins were minted in vast quantities over a period of 2,000 years - literally millions or even billions of coins finding their way into circulation.

Early Chinese coinage is a complex specialised field of study but there is a great deal of detailed information about the coins provided, on the Internet, by the experts.
In reviewing the available data, surviving coins minted from A.D. 1055 (Liao Dynasty) until the 15th C (during the Ming Dynasty) were included - on the assumption that earlier minted coins remained in circulation for several hundred years so might have still been available at the time 'Taiyin Daquanji' was written.

The Ming Dynasty (beginning in A.D. 1368 ) saw the introduction of new standards in the minting of coins. In particular coin denominations of 1 cash, 2 cash, 3 cash, 5 cash and 10 cash were specified proportionally by weight - for example, a 3 cash coin should weigh 3 times more than a 1 cash coin, a 10 cash coin 10 times more and so on. It was specified that 160 of the 1 cash denomination coins should be made from 1 jin (equivalent to 604.79 grams) of copper. As the coin value was based upon weight, the coins were called qian so that 1 qian (1 cash) weighed 3.78 grams. In practice - because of dimensional tolerance variations inherent in the casting process and general lack of precise control over the minting of the coins - quite wide weight and dimensional variations are to be found in the surviving coins from this time period.

The best approach, therefore, has been to determine an average value for coin weights and dimensions of the period.
The attached plot of the weight of 1 cash, cast bronze coins is derived from the measurements of 124 specimen coins. Coin diameters of this sample range from 23.5 mm to 25.5 mm.
From this data, the average coin weight was calculated to be 3.7 grams with an average diameter of 24.5 mm. The average coin thickness was calculated to be around 1 mm assuming an alloy density of 8 grams/c.c.

So, from this it might be concluded that - according to the 'Taiyin Daquanji' - the weight required to correctly load silk 'threads' during the twisting process is about 15 grams per thread (i.e. the weight of 4 coins of 1 cash denomination) - not including the weight of the wooden frame of the 'Zhui'.
How can we know that 1 cash coins were intended to be used rather than the higher denomination coins of 2, 3, 5 or 10 cash? Turning again to John Thompson's translation - the original author of the text adds that one coin may take the place of three. If interpretation of the text is correct, therefore, this might only mean that a 3 cash denomination coin may be substituted for 3 coins of 1 cash denomination. Therefore, the 'small coins' used for the weights would usually be 1 cash
(1 qian) coins.

Turning to the translation of the 19th C text 'Yuguzhai Qinpu' for comparison, the weight specified for loading the strings while being twisted is given as 4 ounces. Assuming the ounce to be a Chinese ounce (tael) which is equivalent to 37.8 grams the total loading would be 151 grams. However, as the text does not relate this loading to any string size or 'thread' count the information can be of no use for this project.

A proper understanding of how a 'Zhui' was designed and used to make instrument strings has eluded me for a number of years as the translations of the early texts are not clear.

I had visions of the string maker laboriously turning the device by hand and counting the several hundred if not thousand revolutions necessary to make a string. Not an efficient method it would seem.
However, a key part of a'Zhui' may be the handle that I have assumed projects underneath the base plate - a superfluous feature it would seem as it would be just as easy to turn the frame by hand.
Another important clue is given under the heading 'Method of Operation' in John Thompson's translation - "If you wrapped (zio: lit rub palms together) (the 'thread' bundles) to the left, then you should combine the strands to the right. The tighter these are twisted together the better". This describes the usual construction of a four 'strand' cord or string.
However, it is the reference to "rubbing palms together" that is interesting. If the handle of a 'Zhui' is rotated by rolling it between the palms of each hand, it becomes in essence a geared friction drive. The attached images illustrate the process (using a pen to simulate a handle). A right handed person gives the handle a rotation to the left (anti clockwise) the number of revolutions per stroke of the hands being dependent upon the diameter of the handle. (I counted four rotations of the pen per stroke and six using a pencil of 7 mm diameter). Additional revolutions would follow given this initial spin due to the weight of the 'Zhui' acting like a flywheel.

There is no mention of the number of turns required to make strings in the early texts only that the 'strands' should be twisted together as tightly as possible. Using this method, however, it is likely that an experienced and skillful string maker might easily judge the degree of twist not only from the appearance of a 'strand' and the increasing resistance to turning the 'Zhui' as twisting progressed but also by keeping each 'Zhui' exactly in line with its neighbours as twisting progressed to completion.
This judgment would also be made easier by making long lengths of string at one time - the vertical movement of each 'Zhui' then being less sensitive to the number of twists than it would be for shorter lengths.

We know from the texts that the early Chinese made their strings in long lengths. More on this to follow.

Both the 15th C 'Taiyin Daquanji' and the 19th C 'Yuguzhai Qinpu' describe the silk strings being twisted in long lengths one end of the string being attached to a tall structure such as a tower and hung vertically with the 'Zhui' at the bottom end.

John Thompson gives the initial length of the 'thread bundles' as 94 ft (29 metres), however, he appears to have used modern Chinese linear measurements in his translation with a Chinese 'zhang' taken to measure 141 inches (this might also be an error as a modern 'zhang' measures 131 inches).
During the Ming Dynasty a Chinese foot (chi) measured about 0.32 metre and a 'bu' (5 chi) measuring about 1.6 metres. On this basis, a zhang (10 chi) would have been the equivalent of 3.2 metres (compared to 3.4 metres today - source Wikipedia).
Therefore, the initial length of the 'threads' for a Ming Dynasty silk string maker would have measured 10 chi X 8 equivalent to 24.8 metres (80 ft) by today's standards. After being twisted into a string, the final length was 25% less or 10 chi X 6 equivalent to 19.2 metres (62 ft).
These corrected lengths now closely match those given in the 'Yuguzhai Qinpu' translation of 80 ft and 60 ft respectively.
The 60 ft length would then be cut into 10 - 6 ft lengths for use as instrument strings.

This string making process, being a vertical rather than horizontal, creates another puzzle. If the twisted string shrinks in length by 20 ft (6 metres), how does the string maker manage to reach the 'Zhui' as the twisting progresses? He might use a spiral staircase 20 ft high perhaps? Another solution might be to design the upper anchor block so that it can be drawn downwards as twisting progresses - as shown in the attached schematic. Nevertheless, the whole apparatus would have to take up space well over 100 ft (30 metres) high.
European string makers made their cords, ropes and strings etc. horizontally using geared 'twirls' mounted on a sled that slid along the ground as the cord reduced in length and provided the required tension.

It will be interesting to test the Chinese method although the tallest building that I have on my property will only allow a finished string length of about 15 ft (4.6 metres) to be made.
More on this at a later date.

The ancient Chinese zither (qin) strings were longer and of heavier gauge than required for an oud and were of four strand (but sometimes three strand) cord construction. Interestingly, the 'Yuguzhai Qinpu' gives the size of the top three treble strings for a medium sized, seven string qin as having 48 threads, 54 threads and 64 threads respectively.

Our project oud strings will have a thread count of 48 for the 'Mathna' and 64 threads for the 'Bam' strings - as specified by the 'Kanz al-Tuhaf'. This is a good indication that these two strings should be made with either a three or four strand construction.
My guess is that the optimum construction of the oud strings might turn out to be:
'Hadd' - 2 cord (2X8 threads = 16)
'Zir' - 3 cord (3X8 threads, or 2X12 threads = 24)
'Mathna' - 4 cord (4X8 threads = 32)
'Mathlath' - 4 cord (4X12 threads = 48)
'Bam' - 4 cord (4X16 threads = 64)

However this is just guesswork at this point in time.

I seem to have omitted the second part of the twisting operation for the Chinese string making process.

Once each of the four 'thread' bundles have been fully twisted into 'strands', the four 'Zhui' are tied together as a group and then released to revolve freely of their own accord. They will then rotate together - under the influence of gravity - in the opposite direction to the twist of the strands (i.e. to the right or clock-wise) combining the strands into a stable string. In order for the completed string to be tightly twisted, the 'strands' must themselves first be twisted as tightly as possible.
This is explained in the 'Yuguzhai Qinpu' as " Revolve it (i.e. the 'Zhui' ) to the left and twist the strands extremely tight. Let it turn to the right and come together becoming a string approximately 60 ft in length" and in the 'Taiyin Daquanji " as "If you twisted the 'thread' bundles to the left, then you should combine the 'strands' to the right. The tighter these are twisted together the better. The original length of eight zhang becomes a string about six zhang in length"

Included here for completeness, the centuries old European 'horizontal' method for making strings, cords and ropes is more complicated (although more conveniently undertaken at ground level).

The attached images from the 18th C. 'L'Encyclopedie' by Denis Diderot (Vol. III, Planche 1, 'Corderie') illustrates the apparatus and procedure for making small diameter two or three strand strings, cords or ropes. The engraving of a rope maker's workshop shows a two strand cord being combined (or 'laid') into a cord - the required tension being provided by a free hanging weight (with the cord passing over a fork) and the twist being controlled using a two grooved 'top' (item c, 'toupin').
Tension is provided when combining a three strand cord (or 'merlin') by a workman holding a swivel hook (l'emerillon) that allows the strands to freely rotate and combine together. (In this case, the 'top' has three grooves).

The 'thread' bundles are twisted into 'strands' simultaneously and equally using a geared machine driving four hooks or 'twirls'. The machine can therefore be used to twist two, three or four strands simultaneously as required.

Enough already - lets try to make some strings!

My vertical rig for the string making trials (images previously posted 28 January) is a compromise taking advantage of the ancient Chinese method of applying a constant tension but using a geared drive to twist the strings. The geared drive will enable an investigation of the optimum number of turns for making each string.

First, a couple of tests were run to verify the functionality of the rig and to get a preliminary 'feel' for the best method. For this test ordinary spun polyester sewing thread was used - itself a mini two strand twisted yarn - so not properly representative of the raw silk 'thread' that will be used for the trials.
As the rig has only one geared drive, each silk 'thread' bundle will have to be individually twisted into a 'strand' then removed from the rig and stored until the remainder of the 'strands' have been prepared. The separate 'strands' will then be attached to a common anchor point so that they can be 'laid' - under direct gravitational loading - into the completed string.
The rig has been provided with a series of peg holes on one side so that the 'thread' bundles can be assembled. This is done by first winding the 'threads' around two wooden dowels spaced a suitable distance apart (e.g. 8 winds for a 16 thread strand). The loose ends are tied off at one end and loops formed at each end by tying the threads together close to the dowels. This then enables the thread bundle to be removed from the rig and easily handled without danger of the threads becoming tangled.
An 8 thread bundle, 1 metre in length, was then placed in the rig for twisting into a 'strand', one dowel being fitted into the double hook on the geared drive and the other dowel into the sliding carriage. The weight box on the carriage was (arbitrarily) loaded to about
1 Kg weight to tension the thread bundle and the bundle then twisted 350 turns. The strand was then removed and stored under tension by fitting the dowels into the peg holes at the side of the rig. A second strand was then prepared and the two strands fitted to a common anchor point. A 1 kg weight was then hooked over the two dowels at the other end and the whole assembly was allowed to freely rotate under gravity to combine the strands together into a string.
The attached images show the completed string - nowhere near the degree of twist required but a useful first step in the 'learning curve'.

A second test was performed with a six thread bundle,
1 metre in length. The 1 kg load caused the bundle to intially stretch by 3 cm. 700 turns were then applied to the bundle , this degree of twisting causing a reduction in length of the strand from 100 cm to 83 cm. Still more turns required!

Next time around, the dowels will be replaced by metal 'S' hooks for the twisting process. The sliding carriage will also be scrapped and replaced by (equal) individual weights for each strand in order to maintain a proper, balanced strand tension throughout.

So far so good. Lets try some silk 'threads' next.

For the preliminary string making trials I shall be using a small sample of raw silk that I have to hand (not the sample material recently supplied by 'Treenway Silks' that has yet to be processed once the necessary skein winding apparatus has been completed).
As I am not sure about the thread size of the silk (it is too fine and soft be measured with a micrometer) a short test string was twisted from 16 threads and the diameter of the assembly measured using a micrometer and dial calipers as 'go/ no go' gauges (i.e. the gap on the measuring tool is set so that it will just slip over the string without pressure being applied)
The diameter measured was 0.018 inch (0.46 mm) so this is a good indication that this silk thread will produce a 16 thread, simply twisted 'Hadd' string of around 0.4 mm in diameter - about where we want to be.
The short test string was then destructively tested by applying tension with a spring balance. The string broke at around
3.5 Kg loading. However, the string would have failed at a higher load had the threads been bound together with gum as a uniform assembly, so this test is a rough indication that a string of this size should be able to work as an instrument string with a tension of 3.5 to 4 Kg.

If my assumptions and calculations concerning the ancient Chinese string making practices are valid (and apply equally to 14 th C oud string making), the thread bundles must be loaded with weights during the twisting into strands as follows:

8 thread bundle - 120 grams
12 thread bundle - 180 grams
16 thread bundle - 240 grams

These loadings apply to strings made from four strands (sometimes three) so may not be valid for smaller diameter oud strings that might have to be made as simply twisted strands (for strength). Nevertheless it is a starting point for these trials. (Note that the loading for a 16 thread bundle is only about 7% of the breaking load of the short test string - which seems to be rather low?))

To tension each thread bundle (8, 12 or 16 in number) at the required load, standard steel washers are to be used (a 3/8 washer weighing 6 grams) placed on a wooden carrier (for each thread bundle).
Wire hooks have been quickly and easily formed from galvanised 14 gauge fencing wire using pliers and a wire bending tool (Lee Valley cat# 92W67.01 @ $8.50).
To prevent the load rotating during the twisting operation, the wooden weight carrier is square based and guided with an adjustable guide. The rig is a bit 'rough and ready' but should be good enough to obtain the data required.

A sample of Gum Arabic has been made up to a very thick, viscous consistency so that addition of a 'little essence of Saffron' will dilute the gum to a workable "paste of moderate consistency" - although at this point it is not known - in practical terms - how much 'a little essence of Saffron" might be (or its concentration) or what is meant by "a paste of moderate consistency" in the Kanz al-Tuhaf instructions for silk string making.

The viscosity of the Gum Arabic sample is such that when the containing glass jar is turned on its side, it takes several seconds before the gum flows under influence of gravity (room temperature 24 C - or 72 F, Relative Humidity 45%). This volume of Gum (10 cc) was then diluted with 0.5 cc of a prepared sample of 'essence of Saffron' (0.25 gram in 10 cc of 40% ethanol in water solution). At this dilution the rate of viscous flow of the Gum was judged to have about doubled.

The addition of the 'essence of Saffron' did not change the colour of the Gum significantly. As the 'Essence of Saffron' is added (according to the Kanz al-Tuhaf) to 'stain' the strings this likely is an indication that the test batch of 'essence of Saffron' is not sufficiently concentrated for the purpose. A more concentrated batch will, therefore, be made up for future tests (using 40% ethanol solution rather than methanol - for safety considerations, methanol being a deadly poison).

The gum is applied to the completed strings "... rubbed on the strings with a piece of linen until it has penetrated into all the parts when the string is dried"
The first preliminary trials with application of the Gum solution have been carried out on the two strand trial string previously made from spun polyester thread - just to see what happens.
More to follow.

The two strand string made from spun polyester sewing thread, with its normal 'back twist' of the combined strands, is a stable string when not under tension but which happens to have an 'open' low twist construction. To increase the degree of twist (necessary for an instrument string) the strands would have to be twisted much tighter before being combined (laid).
However another possibility might be to start with low twist strands and then apply extra back twist to the string to increase the degree of twist in the finished string. This results in an unstable string that will combine upon itself once the string tension is released. So, for this to work, the extra back twist would need to be locked in place with gum before tension is released.

The string construction alternatives therefore are:
1) simply twisted - unstable
2) 2, 3 or 4 strand - stable
3) 2, 3 or four strand with additional backtwist - unstable
We do not know what form of construction applied to the 'Kanz al-Tuhaf' oud strings so must try all of the possible alternatives to find the optimum solution.

Application of the gum arabic after a string has been twisted (yet still under tension) serves not only to bind the string into a uniform mass but also to stabilise those strings with unstable configurations.

To put matters to the test, the polyester trial string was cut in half - one half left with the normal back twist and the other with additional back twist applied. Both string sample were placed under tension - to draw the threads tightly together - and the Gum Arabic was rubbed into each string.
The Gum took only a few minutes to dry at room temperature.

The attached before and after macro images show the result. (note that the spun polyester thread is hairy and not suitable for instrument string making. Silk filament will not suffer from this problem).
The macro images show that the Glue contains many very small air bubbles (undesirable) - an indication that the Gum viscosity was too high and needs to be further diluted for future trials.

The strings - after the Gum has dried are a little stiff after tension is released - so are first 'conditioned' by wrapping them around a pencil. This gives the string the flexibility of a bootlace - exactly what is required.

Out of curiosity, each 'string' (35 cm long and 0.71 mm diameter) was brought up to a higher tension and plucked with a fingertip. To my surprise the short thick strings sounded quite sonorous with a sustain of a few seconds.
An encouraging result to start with.

A new batch of 'Essence of Saffron' has been made up for the next set of trials with reeled silk thread.
The first batch was made from 0.25 grams of Saffron in 10 ml (cc) of 40% ethanol solution in water. This concentration failed to stain the Gum Arabic. This may have been because the Saffron was too old and dried out having been stored in our kitchen for many years.
The new batch has been made from recently purchased Saffron, 1 gram being added to 10 ml of 40% ethanol in water solution. Most of the solution has been absorbed by the herb and the golden yellow stain is sufficiently concentrated to colour the sides of the glass container. The liquid also smells strongly of Saffron.
A further 5 ml of 40% ethanol has been added to the new batch so that there is a liquid level above the Saffron in the container. It will likely take a while for all of the dye to be extracted. Further ethanol solution may have to be added later.

The first stage of the trial for making strings from reeled silk will be to prepare two 16 thread simply twisted 'Hadd' strings - one with low twist the other with high twist - for comparison. As well four 8 thread strands will be be prepared two to be made into a two strand 'Hadd' string with normal back twist, the other into a two strand string with additional back twist.

Two samples of 16 thread strand bundles have been prepared ready for twisting trials.

The silk thread used for these preliminary tests under magnification appears to be made from two reeled threads of raw silk loosely twisted together (see attached image) - possibly originally a fine silk embroidery yarn? No matter, it will do for these preliminary trials although the ultimate tensile strength of the completed oud strings will be somewhat lower than if reeled cocoon filament silk had been used.

The strand bundles were wound (8 windings) between two wooden dowels placed a metre apart on the string twisting rig. The ends of the bundles were tied together at each end with a piece of cotton sewing thread and the loose ends of the silk threads knotted together around the dowel to complete the bundle.
The first string bundle contains a small knot where the silk worker has repaired a break in the thread - normal practice for the industry. For future string making, such discontinuities will not be included.

As previously discussed, we know from the description in the Kanz al-Tuhaf that the silk threads were de-gummed (the sericin gum dissolved) before being made into strings by boiling the threads in water to which wood or vegetable ash has been added. This is a fairly critical procedure as soaking silk in too strong a caustic solution for too long at too high a temperature can damage the silk.
For this test the de-gumming solution was made by adding a tablespoonful (15 ml) of wood ash to 1.5 litres of boiling water.

To avoid tangling of the thread bundles during de-greasing, the bundles were wound around a large coffee can. The can was then placed in the hot de-greasing solution (78 C/175 F), and weighed down by adding 1.5 litres of hot water to the can and the silk thread bundles allowed to remain immersed for 10 minutes. The can was then removed and the (hopefully!) de-gummed threads washed under fresh running water. The can and threads are now left to dry thoroughly before being twisted into 'Hadd' strings.

After drying on the coffee can overnight the two degummed thread bundles were removed for twisting into strings.However, the thread bundles were found to be lightly stuck to the can and after being gently removed had the appearance of a flat, loosely bound silk tape.
Clearly the degumming operation had not completely removed the Sericin. Either the concentration of the caustic degumming solution was too weak or immersion time was too short or solution temperature too low or a combination of these factors.
The degumming operation must, therefore, be repeated until all Sericin is removed and the silk threads can be twisted in a dry state.
The ancient Chinese string makers - as we have seen - did not attempt to remove the Sericin - but twisted the threads when wet with the Sericin in a softened state using the Sericin to bind the silk threads into a uniform string.

The Kanz al-Tuhaf on the other hand states that after boiling the silk threads in the caustic solution of water and ash (i.e. degumming) they are then dried - and then presumably twisted into strings in a dry condition (but that is not confirmed in the manuscript). The threads were then bound together with Gum Arabic.

Efficient degumming of silk is a critical operation a technique well understood and practiced by the silk industry. Perhaps it might be easier in the long term to purchase reeled silk already degummed from suppliers (if such a product is commercially available - which I doubt)?

A second attempt was made today at degumming the thread bundles - this time following my own recommendations (!) following tests on wood ash alkalinity previously posted.

The recommendation is 1 tablespoon (15 ml) of wood ash for every 500 ml of water. The raw silk is then simmered for 30 minutes in this solution (simmer temperature is just below the boiling point of water - about 98 C). After simmering the silk is then rinsed thoroughly with clean water to remove all traces of the degumming solution.

In an attempt to prevent tangling during degumming, each bundle was loosely tied with cotton thread to a wire ring. This, however, was only partially successful as some tangling of the threads still occurred due to water movement during simmering. The thread bundles were carefully separated while still wet but the tangled end loops may have to be re-tied before twisting into strings.
A better arrangement would be to wind the thread bundles fairly tightly onto a cylinder keeping each winding separated and with the end loops anchored to pins.

The degumming operation using wood ash solution seems to have been successful, the silk now having a silvery sheen and the threads no longer stuck together. With the sericin gum removed some slight 'hairiness' of the threads can now be seen in places due to separation of the fine cocoon reeled filaments making up the threads.
Interestingly, ash (from the Banana plant) is used today in India to degum the cocoons of the wild Tussah moth.

So - with the thread bundles now degummed - on to the twisting operation.

A silk 'Hadd' string (#1) is made! A 16 strand, degummed thread bundle has been twisted into a single strand 'Hadd' string.

The first objective is to twist a string with the judged minimum number of turns necessary to make a viable, uniform cylindrical string. For the small diameter treble strings tensile strength is more important than elasticity - low twist strings being stronger than the more flexible (but weaker) high twist strings.

The thread bundle was loaded on the twisting rig with 240 grams (previously determined from the ancient Chinese string making data). A total of 175 turns was then applied to give - what was judged to be - the minimum turns required for a cylindrical string. At this point the twisted string - under full load - shortened in length by 17 mm.
After rubbing a Gum Arabic/Saffron dye solution into the twisted string with a piece of cotton cloth - the string then increased in length by 22 mm. The string was then allowed to dry while still under load.

The Saffron dye solution used to dilute the Gum Arabic was made from 1 gram of Saffron herb soaked in 25 ml of 40% ethanol in water. The dye colored the gum to a rich golden colour. The gum dilution was such that it flowed rather like the consistency of hot hide glue.
The dye enabled gum penetration/coverage of the string to be verified.

The result (#1 Hadd) is a smooth golden coloured string measuring 0.38 mm (0.015 inch) in diameter with a twist angle of about 10 degrees (low twist).
However, a macro examination of the string reveals a 'hairy' surface due to the loose microscopic degummed fibroin filaments. I thought that wiping the twisted string with Gum Arabic would cement any loose filaments of silk together but the friction of the rubbing procedure likely has produced an even greater degree of 'hairiness'. I shall experiment later on to see if it is possible if the 'hairiness' can be smoothed out mechanically.
The little repair knot previously noted is still in evidence as a flaw conforming that all thread bundles should be free of this kind of discontinuity.

Although the degumming procedure has been effective the silk - although silky smooth and easy to twist without friction - then becomes difficult to handle as the silk filaments are no longer held together by the Sericin gum. On thread bundle #2 even the final knot, tying the bundle loops, had come undone during degumming!
Much of the disturbance of the filaments of the threads may have occurred while handling the thread bundles immediately after degumming - so perhaps with the improvements to the degumming set up as previously proposed (and more practice!) this may be less of a problem in future.

Checking again the English translation of the description of the string making process in the Kanz al-Tuhaf there is no question that the threads were degummed before being twisted into strings after which the Gum Arabic/Saffron dye was applied.

The completed string removed from the twisting rig, after the Gum Arabic had dried, was slightly stiff so would have 'kinked' if subjected to tight bending. The string was, therefore 'conditioned' by applying a technique developed by Alexander Rakov. The string is wrapped around a small diameter wooden rod with a single turn and then - with the string under slight tension - the rod is rolled along the length of the string. This causes fine cracks in Gum Arabic coating giving the string the required flexibility.
On string #1 this also caused some loosening of the twisted thread bundles causing the string to unwind a little for a few turns. This is an indication that the Gum Arabic had not penetrated the string completely. So for string #2 a more liquid solution of the gum will be used to allow better absorption into the string (and reduce friction on the string when rubbing in the gum).

Next, thread bundle #2 will be twisted to determine the maximum amount of twist it can take.

Lightly rubbing string #1 between the fingers seems to be effective in removing most of the silk 'hairs' from the string surface.
In doing so, a section at one end of the string - about 2 inches (50mm) long - started to partially unravel. The macro image shows that the Gum Arabic/ Saffron dye had not penetrated to the centre of the string, remaining only on the surface. The rest of the string appears to be intact so the problem may be confined only to this section at the end of the string.

Thank you for your kind words and encouragement Fernand. As a youngster I did read the series of science fiction books "Les Voyages Extrordinaires" by Jules Verne but - for some reason - not the 'sequel' that you refer to. (Perhaps it was not in the public library that I visited at the time).
I eventually gave up my chosen career as an engineer to take up more creative (much more interesting but much less remunerative!) activities later in life.

I am posting details of my 'laboratory notes' on the forum in the hope that others - interested in the early history of the oud (or lute) might be encouraged and enabled to try making silk strings for themselves.

After some hesitation, Hadd string#1 was put to the test today.
The test bed is a lute that I made some years ago - a copy of a late 16th C original by Giovanni Hieber, 7 courses, string length 60 cm currently strung with Pyramid PVF trebles and metal over spun basses. The top course is a single string, the remainder unison tuned.
The lute is pitched in F at A440 i.e. the top string is tuned to f'which is about as high as a gut or silk string might be tuned to avoid frequent breakage. (although up to a tone higher might be possible with more frequent string breakage).
The gut frets on the lute are worn way past their 'replacement overdue' date but will do for this initial test.
It so happens that the the little knot flaw in the string when under full tension has moved just beyond the nut - out of harms way.
The strings are plucked with fingertips (no nails!) using an historically correct 16th C 'thumb out' technique (The shape of my thumb nail prevents me using the alternative 'thumb under' technique of the period).

The string is still 'settling in' but I am quite surprised at the sound which is as clear and resonant as the PVF strings but with a more complex 'woody' sound - hard to describe. So here is a brief mp3 clip of the string sound recorded with a Zoom H2 digital recorder.
If I can figure out how to compress the audio file below 1 Mb, will post a clip of a lute piece showing how the string performs together with the other strings.







Attachment: STE-006 clip.mp3 (450kB)
This file has been downloaded 521 times

Thanks Dave H. The first result is certainly encouraging.
Note that the sound files have been compressed by about 10 times so some acoustic detail has been lost in the conversion (i.e. it sounds even better in the original recording). Also note that the files are 'raw' data as recorded by the Zoom H2 digital recorder with no editing or effects added. The recorder was positioned above the lute at about 2-3 ft distance away.
Replacement of the badly worn frets on the test bed lute will no doubt make another big improvement in the sound.

I do not have reliable comparative data on tensile strength of silk and gut but I would expect silk to be stronger - the ancient writers on the oud certainly thought so preferring silk for the treble strings as they not only sounded better but could be made smaller in diameter than gut.
Silk is said to be stronger than steel (but it depends upon the steel I suppose) - almost equivalent to Kevlar. It is attracting a lot of attention today (together with spider silk that is even stronger) for commercial applications of artificially spun fibroin filament that is more uniform than natural silk.
Silk filament will stretch to 13% of its original length before breaking but has poor elasticity (i.e. once stretched does not recover to its original length). So, like gut bass strings, silk strings must be made more elastic by virtue of their twisted or roped construction - more twist = greater elasticity = lower tensile strength.

The dilemma for modern historical string makers is that the gut or silk available in Medieval or Renaissance times may have been significantly different than is available today - the properties of both materials being dependant upon the strain of the moth or breed of sheep, their feed and environmental conditions etc.

I suspect that the 60 cm low twist Hadd #1 string will go to a higher tension and pitch (g' at A440 perhaps) but I am starting at f' to see how the string holds up to the stress of being played before increasing pitch. All of the experimental strings will be carefully tested to destruction - plenty more where they came from!
After 24 hours at full tension, the Hadd #1 string is still stretching a little but now seems to be stabilising in pitch. So far the string, under full load, has stretched to about 5% of its original length of 60 cm (judging from the movement of the little repair knot). Interestingly, the diameter of the string has not changed under tension from its original 0.015" diameter (0.38 mm).
The gummed string is not sticky in any way under the fingers but has a good tactile 'grip' for fingertip playing (unlike plain nylon) and has a brightness equivalent to PVF but a more interesting tone colour (in my opinion).
It remains to be seen how well the gum coating will wear in service. However, as the gum is water soluble, it may be possible to maintain a string by wiping off the worn gum with a hot moist rag and then rejuvinate the string with a new coating of gum. Something to experiment with.

For information I have attached another sound clip of the string sounded 'open' and as far as the 12th fret position. The test bed lute has only 8 frets to the end of the fingerboard (typical of lutes of the period). Lutes of the period were played above the 8th fret onto the sound board without use of frets. The early lutes had this area of the sound board made more durable with a hard varnish coating as can be seen in the attached image of a painting of a lutenist (Francesco da Milano?) dating from the first half of the 16th C.

High twist Hadd #2 has been made - more to follow.


Attachment: Hadd 1 clip.mp3 (257kB)
This file has been downloaded 518 times

" A paste of moderate consistency is then made of the gum and a little essence of saffron. This is rubbed on the strings with a piece of linen until it has penetrated into all the parts, when the string is dried" (Kanz al-Tuhaf)

Although the Hadd #1 string appears to be uniformly and smoothly coated with Gum Arabic, unraveling a section of the string confirms that only the outer threads have been coated with the gum. Indeed, it would seem to be impossible (from this one experience) for the gum to completely saturate the relatively tightly wound string to its core (if that is what 'penetrated into all the parts' is taken to mean).
The attached macro image shows that a majority of the threads in the string remain free of gum (note the now sericin free microscopic fibroin filaments - confirming the efficiency of the initial degumming procedure). The Gum Arabic in this case is, therefore, a surface coating that fills only the external irregularities of the string.

To test if complete gum penetration is possible the Gum Arabic will be further diluted before being applied to Hadd #2 string. Also instead of using a piece of fine cloth (which absorbs the gum) the gum will be rubbed in with the fingers - wearing non absorbent vinyl gloves.

Hadd #2 - made from a 16 thread bundle like Hadd #1, loaded with a 240 gram weight, was twisted 525 turns (compared to 175 turns for Hadd #1). This produced an angle of twist in the threads of about 45 degrees (compared to 10 degrees for Hadd #1 - see attached image) and a reduction in length of 13 cm from the original untwisted length of 100 cm. The string diameter increased to 0.41 mm (0.016 inch) - compared to 0.38 mm diameter for Hadd #1.
This seemed to be an extreme angle of twist so twisting was stopped at this point - not wanting to test this string to destruction.

The Gum Arabic solution was further diluted with the addition of 5 ml of 40% vodka making a fairly fluid solution but with a less intense dye colour. This was rubbed into the string with gloved fingers - several applications being made. However, this resulted in an unsatisfactory coating that broke apart due to rubbing in the gum as it became 'tacky' or partly dry. The advantage of using a cloth pad to apply the gum is that the cloth absorbs some of the liquid gum - acting as a reservoir and so slowing down drying of the gum. So, we live and learn - and no doubt the ancients would be saying 'I told you so"!

Rather than scrap this string, a cloth pad was moistened with hot water and rubbed over the string which quickly and effectively removed the crusty gum layer. The cleaned string was then successfully recoated with gum - this time using a slightly damp cloth pad to reduce absorption of the gum into the pad. The less intense dye colour made coverage of the string with the gum a little more difficult to judge.
The larger broken silk filaments seen in the cleaned string may be an indication that a maximum degree of twist has been reached but is more likely to be due to mechanical damage during the first, unsuccessful, gum application.
The short black fibres seen in the images may be foreign material attached to the raw silk as well as from the cotton cloth used to apply the gum. Some small black particles were also in evidence that likely came from the wood ash used for degumming the raw silk.

The slight surface irregularities seen in the surface of the re-coated string will likely smooth out once the string is brought up to an estimated full tension of around 3.0Kg (f') or a possible 3.7Kg (g').

The next test will be to twist a 16 thread bundle to destruction - to determine the maximum degree of twist possible.

Improvements to be made for the next string:-
1) Filter the wood ash from the degumming solution so that the strings boil in a clear caustic solution.
2) Make a copper cylinder to hold each thread bundle securely during degumming to prevent tangling and mechanical damage to the threads.
3) Increase the viscosity of Gum Arabic by evaporation and then dilute again with the concentrated 'essence of saffron' solution - to increase the intensity of the dye colour to the original golden yellow. This will help in judging gum coverage of a string.
4) I have a stock of very fine linen cloth salvaged from old engineering drawings. This will be used in place of the cotton cloth for application of gum - for greater 'authenticity'.

After 4 days of playing Hadd #1 on the test-bed lute at f' (A440) pitch, one of the threads of the 16 thread string broke at the 5th fret. The damage is mechanical - due to the worn gut fret cutting into the string so is not a tensile failure.
Some 'hairiness' of the string was also seen - under magnification - just below the rosette where the string is plucked. This is (less severe) mechanical damage - the 'hairiness' being due to broken microscopic Fibroin filaments appearing on the surface of the string.

As a test, the broken thread was cut away on either side of the fret position and the string was re-coated with fresh Gum Arabic. This sealed the broken thread securely in place as well as any 'hairiness'.
With this string maintenance completed testing continued for another 24 hours when another thread broke - again at the 5th fret.
Macro examination of fret #5 reveals worn, sharp edges (the string has been pulled to one side for clarity in the image). This wear of the fret would not have damaged a nylon or PVF string (more likely it is these strings that caused the fret damage in the first place - the nylon/PVF being harder than gut).

The test of Hadd #1 has, therefore, been terminated at this point. The test bed lute will now be re-fretted with new gut (long overdue!) before proceeding with a test of Hadd #2.
The new fretting will be double strand (rather than the current single strand of gut). A double fret wrapping is more historically appropriate and has the advantage of increasing fret useful life (the wear being shared over two gut strands rather than one) and enables the frets to be tied tighter to the neck/fingerboard.

It is interesting to contemplate why the Medieval Persian string makers chose a different path to that of the Chinese - preferring to de-gum the silk threads before twisting - and then applying Gum Arabic as a binder - rather than working with the silk's natural gum Sericin. De-gumming the silk is an additional procedural complication and makes handling of the silk threads more difficult. It is unlikely that the Persians were unaware of how the Chinese made their strings (the string making information was, after all, published in printed Chinese books of the period so was not a secret).
There is a significant tone colour difference between strings made from raw silk and those made from de-gummed silk (ref. Alexander Rakov) - so maybe it was just a matter of auditor preference . However, use of Gum Arabic to bind the string threads together in place of the Sericin does have a potential advantage in allowing the string life to be easily extended by re-coating with gum as necessary. Perhaps this may have been a factor?
There is no information to confirm that the de-gumming procedure of the Kanz al-Tuhaf was thorough enough to remove all traces of Sericin or only (most likely) partially so. So we must 'remain in the dark' on that question.

Silk strings made the old Persian way, would seem to be sensitive to mechanical failure due to abrasion (based upon our extensive experience with Hadd #1!). Perhaps this is one reason why frets were eventually abandoned on the oud - because they contributed to excessive string wear? Just speculating! Note , however, that the thin Hadd trebles will be more susceptible to failure than the more substantial basses - higher tension, smaller diameter.

The 'test bed' lute has now been re-fretted. It will take a while for the frets to settle in and be fine tune adjusted.
Hadd #2 has been fitted and is still being stretched to full tension (f' at A440) so it is early days to judge. First impressions are that the string has a bright sound and tone colour (like Hadd #1) but feels a little too 'stretchy' when being plucked. This might be explained by the fact that Hadd #2 is a high twist string (higher twist = greater elasticity).

Back to the question of Gum Arabic.
"A paste of moderate consistency ..." (Kanz al-Tuhaf).
At the outset I have assumed that 'paste', in this context, means 'a moist, fairly stiff mixture'. However, from the initial tests it is clear that if the Gum Arabic is made too stiff or viscous it will not 'penetrate into all the parts' of a tightly twisted string - no matter how hard it is rubbed in.
Another meaning of 'paste' is a relatively weak adhesive used to stick paper together (e.g flour and water paste). So -should the translator of the Kanz al-Tuhaf text more accurately have used the word 'glue' instead of 'paste'?
Gum Arabic can be used as a glue for paper (when suitably diluted - in order to make it flow so that it is easy to apply with a small brush). I imagine then that Gum Arabic, made as a glue, would have had a 'watery', varnish like fluidity - similar to the liquid mucilage glue that was once commonly used in schools, in offices and around the home until the early 1960's. One popular brand was LePage's mucilage glue that came in a fancy bell shaped bottle with a red rubber 'Gripspreader' cap. I remember it well but it seems that this product is no longer available having been replaced in recent years by synthetic glues (but probably these products are still called 'mucilage glue').
Strictly speaking, vegetable mucilage is not identical to natural vegetable gum but is similar to the point where the terms 'mucilage' and 'gum' are interchangeable in common usage when referring to glue.
I think that I can still recall how LePage's mucilage glue flowed from the bottle so will further dilute my Gum Arabic solution accordingly with the 'essence of saffron' - to try on the next experimental string. Hopefully this will result in a better penetration of the gum into the string.

Hadd #2 string broke today at full tension (f' @A440) after only 20 hours and before stretching fully. The break was at the front edge of the nut - this being the point of highest stress for the string.
Although Hadd #2 is less than perfect, the reduced service life (compared to Hadd #1) is not altogether unexpected due to its high twist construction.
Hadd #1 was made with a 10 degree angle of twist. So for a string tension of 3 Kg, the twisted threads in the string would have to sustain a tensile load of about 3.1 Kg. By comparison, for the Hadd #2 string, with a 45 degree angle of twist, the twisted threads would need to sustain a tensile load of about 4.2 Kg. This stress load is further increased at the nut due to the stress concentration in that location - clearly too much for a string of this dimension and structure to sustain. So no surprises here - confirming that a simply twisted Hadd string should be made with as low a twist as is practical.

In order to gain as much information as possible from the failed Hadd #1 string, a section was coated with a thin rubbing varnish ("Minwax - Antique Oil Finish"). The result is a smooth flexible string that may have better abrasion resistance than Gum Arabic? Something to investigate further.
The flexibility of historical strings (gut or silk) should allow them to be tied like 'bundles of shoe laces' as is Mimmo Peruffo's (Aquila Strings) apt description.

Four thread bundles - 0ne 16 thread and three 8 thread have been prepared for testing. The 16 thread bundle will be tested to determine the maximum degree of twist that it can sustain. One of the 8 thread bundles will be similarly tested. The remaining two 8 thread bundles will then be twisted into a 2 strand Hadd string - just to see what happens.

To de-gum the thread bundle samples, they have been wound on to a 'de-gumming' tube - to keep the thread bundles separate and under control to ease handling.
The tube has been made from a 4X4 ABS plastic sewer pipe connector (available from any hardware store). The surface of the connector was first inspected for any rough spots that might break the silk threads. These were smoothed off with fine sand paper. Self tapping screws set in the connector serve as anchor points for the thread bundles - tied with string to the anchors.

The de-gumming solution was made by dissolving wood ash in water (25 grams / litre). The wood ash was wrapped in a fine cotton cloth (to keep the solution clear of ash debris) before being immersed in water that was then brought to the boil. The PH of the solution, after a few minutes of boiling, reached about 10 (tested with PH indicator papers) The 'degumming tube' was then fully immersed in this solution and allowed to simmer for 30 minutes. The tube was then removed, thoroughly rinsed in clean water, and set aside to dry.

The degummed 16 strand and 8 strand bundles have been twisted to destruction to determine the maximum number of turns that can be carried by each. This information is required for multistrand string construction where each strand (or thread bundle) must first be twisted to the maximum degree before being combined (or 'laid') together.
How is the maximum degree of twist determined? It is the point - before the threads break - where the thread bundle is so tightly twisted that it begins to take up a distorted 'wave' form - from the desired uniform cylinder.
For the 1 metre long 16 thread bundle - loaded at 240 grams - this point was judged to occur at about 550 turns. This thread bundle broke at one of the end loops at just over 800 turns.
For the 1 metre long 8 thread bundle - loaded at 120 grams - twisting continued until the thread bundle broke at about 1150 turns. However, maximum twist was judged to occur at about 1000 turns (by examination of the bundle under magnification as twisting progressed).
The difference between the results for the two bundles can be explained by the diameter difference of the twisted bundles. The threads in the 16 thread bundle are under greater tension than those in the smaller diameter 8 thread bundle.
The attached images show the graphical plots of the contraction in length of each thread bundle against the number of turns - measured on the test rig as twisting progressed.

The broken 16 strand thread bundle was allowed to fully unwind (with a small load attached to one end) and was then re-twisted to the low twist limit of 175 turns. The string was then coated with gum so that it could be tried on the lute test bed.
This time, fine linen cloth was used to rub a more dilute Gum Arabic solution into the string. Application of gum with a dry linen cloth produced an uneven coating. This was wiped away with a cloth soaked in hot water and a new gum application made - this time with the linen cloth first soaked in water - which produced an even coating of gum.
This string - Hadd #1A (low twist)- is now under trial on the test bed lute.

Just for clarification - DaveH commented in his previous post about string 'flexibility' although he probably meant string 'elasticity'?

An instrument string must have sufficient elasticity to function well - i.e. if it is stretched slightly (within its elastic limit) it will return to its original length.
All elastic instrument strings are flexible (pliable - can be bent) to a certain degree. However, early lute strings demonstrated extreme flexibility in that they were sold in long lengths tied up like bundles of shoe laces (known as 'knots').
The attached image of a 16th C engraving of a lute string 'knot' demonstrates this extreme degree of flexibility.
Hans Holbein depicted an image of a lute, with seemingly photographic accuracy and detail, in his famous painting "The Ambassadors", 1533. The lute has a broken octave treble string on the fourth course - seen below the pegbox and at the bridge. Note how flexible this string appears to be.
Compare this to a modern plain gut instrument string - supplied carefully wound (in short lengths) in a coil to avoid any sharp bends or 'kinks' that might spoil the string. This is an example of low string flexibility (but probably a string that may have adequate elasticity).
Modern (20th C) Chinese silk instrument strings are of four strand roped construction. These are flexible strings but supplied (probably for convenience or by tradition) in coils. The length of string in the image is about 10 metres.

The objective of these trials is to make silk strings with both extreme flexibility and adequate elasticity to perform well acoustically. Durability of a string is another important objective - although lack of durability of the thinnest treble strings (under the highest tension) has always been problematic.
We do not know, however, how flexible the early oud strings were. My hunch is that they were as flexible as the early lute strings apparently were - be they gut or silk.

The two 8 thread bundles for Hadd #3 were first twisted separately on the rig to 875 turns each - lower than the 1000 turn maximum twist limit previously established in order to provide a factor of safety.
The twisted bundles were then attached to a hook together with the 120 gram weights being careful not to allow the tension to relax. The two strands were then together - top and bottom - with some cotton thread and the whole assembly allowed to rotate freely under the influence of gravity. This resulted in an even, open twisted two strand string of stable construction (i.e. a yarn).
The string was then rubbed with Gum Arabic and allowed to dry.

Interestingly, each strand when twisted to 875 turns reduced in length from 100 cm to 80 cm. After combining the strands, the string length increased to 88 cm and then to 89 cm after applying the gum coating - an 11% reduction in length.
According to the early Chinese string making accounts, however, strings made this way should end up by being 40% shorter. This implies that the Chinese must have added additional turns to the combined strands - in order to create a smoother string perhaps (although the early texts make no mention of this additional step in the string making process).
Note that the Chinese strings were usually of four strand construction (but sometimes 3) never two strand - presumably in order to make a 'smoother' string

The open twist construction of Hadd #3 would seem to make it an unlikely candidate for a treble string because, although it is uniform, it is not a uniform cylinder (although the cross section does come close to being circular). Nevertheless, the string has now been put to the test on the lute.
Note that plucked, wire strung fretted instruments of the 16th/17th C - like the orpharion and bandora - used brass strings of this two strand construction for their basses.

Hadd #3 test results to follow.

The Hadd #3 string felt a little unfamiliar under the fingers on test - not being 'smooth' to the touch (compared to a monofilament PVF string) and more flexible. Plucking the string was more difficult to control, the finger tips tending to 'grab' the open twist causing slight string "clatter" on the frets. This can be heard on the attached audio clip - although, overall, the sound was judged to be bright and 'silvery'. The extra flexibility also has the advantage of providing a clear response when stopping the string beyond the frets on the lute.

This string stretched 8% in length when initially brought up to pitch (f'@A440). After 43 hours in service the string failed while being tuned up to pitch - the failure again occurring just behind the nut (one of the two strands broke).
It is clear, from the tests so far (as anticipated), that the best string construction for the thin Hadd treble is a simply twisted string with a low angle of twist - this being the strongest configuration.

The next step will be to jump to the 'other end of the spectrum' and make a simply twisted 64 thread Bamm string for comparison before going too far into multi strand string construction.
For this trial a low, medium and high twist string will be made and tested. A 64 thread bundle will first be tested to destruction to determine the maximum number of twists it can take. I don't like breaking strings but, after all, one has to 'break eggs to make an omelette'!




Attachment: Hadd 3 test.mp3 (676kB)
This file has been downloaded 482 times

Winding the silk threads between the two wooden dowels on the twisting rig is slow and inconvenient - not too bad for a 16 thread bundle Hadd string requiring only 8 full turns around the pegs but for a Bamm thread bundle of 64 threads requiring 32 turns, a more efficient method is required.

The 'strand winder' (let's call it that) has been quickly and easily made from scrap materials. Essentially a wooden cruciform winding frame on a support frame with a holder for the silk bobbin. The silk thread is fed through a plastic coated hook to guide the silk thread centrally onto the winding frame. The left hand applies light friction to the bobbin to maintain constant slight tension to the thread during winding while the right hand rotates the winding frame.
Three removable peg positions have been provided on the winding frame to give strand lengths of 100, 110, and 120 cm
as required dependent upon string reduction in length after twisting (at present unknown for a silk Bamm string).

Once the required number of turns have been made, the thread ends are tied and the thread bundle or strand then transferred to the twisting rig.

I attach revised test graphs (previously posted) for establishing experimentally the maximum twist for 8 and 16 thread bundles. These are now named Plot A and Plot B respectively.
The revision superimposes - for comparison and information - optimum minimum and maximum string twist derived from a formula by Alexander Rakov established during the course of his development work in making simply twisted silk strings using the Chinese method of using reeled raw silk filament (thread) rather than - in our case - degummed silk.
Alexander found that the optimum number of turns needed to make workable strings could be calculated according to the following relationship:
Number of turns (N) = Co X Length of thread bundle (untwisted) divided by the square root of the number of individual filaments in a bundle.
The Coefficient of Twist (Co) varying between 25 and 37.6.

It should be noted that in our case, each thread is comprised of two reeled filaments.

In Plot A it can be seen that the maximum twist arrived at by experiment (1000 turns) agrees closely with the Rakov calculated value. (In this experiment the minimum twist value was of no interest). Interestingly the maximum and minimum twist values lie over that central portion of the graphical plot that is essentially a straight line.

In plot B, the Rakov values again lie pretty well on the central straight line section of the graphical plot.
The twisting rig experiment established a high twist value for the Hadd #2 string of 525 turns which falls midway between the Rakov calculated values.
However, the Hadd #1 minimum possible twist that was established (for maximum strength) - judged experimentally at 175 turns - is a significantly lower value than the calculated minimum of 439 turns.

The degumming tube has been modified in an effort to make the degumming procedure more efficient. Small plastic tubes have been fitted to the main body so that one end loop of a thread bundle can be attached to one tube, wrapped around the main body and the other end loop attached to the nearest tube with a thin elastic band (to keep each thread bundle in place under slight tension).
Unfortunately with this arrangement a 16 thread bundle and 64 thread bundle on test did not fully degum - the threads being glued together to make a silken tape together with small patches of sericin where the elastic bands had come into contact with the threads during the degumming procedure. Furthermore, the 16 thread bundle had snapped under the tension of the elastic band.
So - back to the drawing board! Degumming of the strands is not going to be easy if complete removal of the sericin is to be achieved without significant loss of strength of the silk threads.

Nevertheless, rather than waste the partially degummed thread bundles it was decided to proceed to subject these failed samples to maximum twist testing - just to see what might be learned.

On the twisting rig it became immediately apparent that the sericin residues were preventing each thread bundle from being smoothly twisted - producing a rather 'bumpy' string. It should be noted, however, that the thread bundles were being twisted dry whereas if the string had been moistened first this would have softened the sericin sufficiently to allow the threads to slide over each other to make a smoother string (this is the old Chinese method for making strings from raw silk)

The 16 thread bundle was loaded with a 240 gram weight and the 64 thread bundle with a 960 gram weight (4X240). As twisting progressed there came a point when the tension of twisting in each bundle overcame the weight load the bundle then suddenly winding around itself like a spring. Lets call this 'corkscrewing' for want of a better term. This is the point of maximum possible twist.
The attached image shows the start of corkscrewing in the 16 thread bundle. Twisting then proceeded further until the thread bundle eventually broke under the strain. The 64 thread bundle was tested in the same manner until it became impossible to control further twisting and take measurements. This thread bundle did not break.
The attached plots C and D show the measured contraction of each bundle against the number of turns. For information, the minimum and maximum twist values calculated using the Rakov formula have been superimposed.
Plot C also includes Plot B for comparison suggesting that the the fully degummed silk thread bundle used for the Plot B trial is significantlyweaker than that of the partially degummed thread bundle. This again suggests that the degumming procedure needs further refinement in order to maximise the strength of the silk threads.

To complete the picture for this part of the trials, the attached image shows the 64 thread bundle in its uniformly 'corkscrewed' state - much shorter than the free length of the original thread bundle. The thread bundle in this state is quite elastic, like a spring, returning to its original state when stretched. The diameter has also increased. I unfortunately forgot to measure the diameter before unwinding the bundle but from the image it can be estimated that the diameter is about 80% greater than the simply twisted (uncorkscrewed string).
Could this construction be made into a viable bass string after being stabilised with a Gum Arabic coating (if the tension is released the string would otherwise unwind of its own accord)? Something to investigate further perhaps.

Allowing the thread bundle to unwind after all this abuse reveals that the silk threads have been permanently distorted beyond their elastic limit - looking a bit like wool from a Merino sheep! The' non woolly' bits are places where the sericin remained after degumming.

Silk supplied as a skein must be mounted on a 'skein winder' so that the silk threads may be unwound without becoming entangled. The loops of the skein - as received from the supplier - are loosely tied and the free ends identified with a short piece of single and double thread.
The silk thread is finer than a human hair and stronger than steel (of the same diameter) - nevertheless careful handling is necessary to avoid breaking the thread.
The 'skein winder' has been made from scrap material. The arms are made from pine and set at a slight angle from the hub so that tension is applied to the string skein supports that may be readily adjusted to accommodate the skein diameter.

The intent is to feed the silk filament directly from the skein winder to the string winder and eliminate the unnecessary step of winding the silk first on to a bobbin.

With the skein mounted on the winder there is some concern about the loose threads remaining 'dangling' after the skein has been fully tensioned. The two free ends of the skein appear to start to tangle in one spot and do not seem to run freely off the winder as expected. However, it is important to select the correct free end - the end that runs from the outside of the skein. Due to twisting of the skein (?), some manipulation of the skein on the winder was first necessary to free the thread from tangling - although it seemed to still temporarily stick in places. As the thread will be fed directly onto the string winder - the whole thing being wound slowly by hand - it is hoped that sufficient control will be maintained to avoid thread breakage.

With the rig set up on the workshop bench, winding a 16 strand 'Hadd' as a test went without a problem. So far so good! With 8 turns of the silk thread on the winder, the two free ends are tied together to complete the thread bundle. Before removing the thread bundle from the winder, two opposite ends are formed into loops (tied with cotton thread) required to hold the retaining hooks during the thread bundle twisting operation.

The skein is labeled with Chinese characters on one side (I don't know what they mean) and the number 40/44 that I take to be the denier measure (weight in grams per 9000 metres). This is given being between 40 and 44 denier - a standard tolerance for the silk industry.
A short, 16 thread, test string was made up and the diameter measured at about 0.37 mm using a dial caliper (difficult to be precise due to the relative softness of the silk). This should be about the required diameter for a 'Hadd' string but will be confirmed, one way or another, after making and testing a few strings. If the string diameter turns out to be a bit too small, then a proportionally greater number of silk threads may be used for each string bundle to make slightly thicker strings.

Under magnification, it appears that the silk thread is 'raw' - that is the Sericin gum has not been removed. The easiest and most efficient procedure would be to de-gum the whole skein prior to making up the string bundles. However, it has been decided - for the purposes of these trials - to de-gum each thread bundle individually prior to twisting. This way will minimise the risk of losing the whole skein should there be thread entanglement during the de-gumming operation.
All a bit slow and tedious but this is not intended to be a production string making process at this stage.

It is the usual practice in the textile dyeing industry to remove the Sericin coating entirely before application of the mordants and dye. This is presumably to ensure maximum penetration of the dye by adsorption by the Fibroin. Degumming of the raw silk is usually accomplished commercially by soaking in a hot alkaline solution with the addition of soap.
However, acid solutions will also dissolve the Sericin (even pure water will remove about 5% by weight). Low strength acids will also remove the sericin without attacking the Fibroin.
Tannic acid (Tannin in solution) may be adsorbed by the Fibroin by up to 25% in a hot solution and by a considerable amount even in a cold solution.
Furthermore, it would appear that it is not necessary to remove the Sericin in order for the mordant salts or acids to be adsorbed (the ancient Chinese even increased the weight of cocoons with salt without need for degumming). One source gives an example of a pound of raw silk that will adsorb half a pound of Tannin without the silk being changed in character and that a pound of silk so loaded with Tannin can then adsorb an additional half a pound of an iron salt without adverse effect. By adding more mordant chemicals, the total weight of a pound of silk might be increased to 3 or 4 pounds (and in some experiments has been increased to 9 pounds!). Silk given a black dye will stand more weighting than silk with a lighter coloured dye.

So the possibility that Ziryab's dyed oud strings may have been weighted with mordant chemicals applied in cold solution is a possibility worth testing as part of this investigation.
A modest doubling of the linear weight of a string will - according to the Mersenne-Taylor law for a vibrating string - lower the frequency of vibration by about 30% (string diameter, length, and tension remaining unchanged). So this is a target weight increase worth aiming for - particularly for the thicker bass strings.

Next - to review the materials required for weighting the silk strings in the context of this investigation where only chemicals that would have been available to silk string makers in the 9th C can be considered.

While searching for the best source of mordants/natural dyes, an experiment to make Stannous Chloride has been tried.

The preparation of the salt is basically straightforward but was found to be somewhat slow and tedious. For convenience lead free solder (95% Tin 5% silver) - available from any hardware store - was cut into small pieces (the smaller the better). Commercial strength (32%) Hydrochloric Acid - available from hardware stores for cleaning stone and brick work - was added (10 ml) to the Tin in a test tube. The acid reacts readily with the Tin to form a solution of Stannous Chloride and Hydrogen gas is given off. Tin was added little by little until no more would dissolve.
To speed up the process, the test tube was immersed in a water bath maintained at a temperature of 90 Celsius. Nevertheless it took several hours to complete the task with about 4 grams of tin being dissolved in 10 ml of acid.
The resulting clear liquid has been stored with excess undissolved tin in case there is any further delayed reaction.
The solution is strongly acid (about pH 1 or 2 measured with litmus test papers) so will have to be used with a buffer solution to reduce the acidity to a level where the silk Fibroin will not be damaged.

As commercially prepared Stannous Chloride in powder form is relatively inexpensive, a small quantity (100 grams) will be purchased for test purposes along with the Tannin and the other chemicals needed for the trials.

Work on silk string making is currently temporarily 'on hold' due to life's other priorities at this time of year.
However, 'Treenway Silks' has a clearance sale and is selling a 200 gram cone of 120/2 spun silk yarn at 50% off regular price for $23.80 (plus shipping and handling). Each cone carries 13,000 yards of silk yarn (about 12 kilometers). I could not resist purchasing a sample for testing!

Spun silk is made by spinning relatively short lengths of silk filaments (from 'scrap' cocoon silk or perforated cocoons) into a thicker strand (unlike reeled silk where a strand is made from a continuous cocoon filament).
Spun silk yarns are made by then twisting the spun strands together - used primarily for weaving into silk fabrics. The industrial standard 120 (e.p.i or ends per inch) means that if 120 yarns are laid side by side they will measure an inch across (25.4 mm). So 120/2 is a relatively fine yarn composed of 2 spun strands twisted together to make the yarn (a stable assembly of silk fibres).
The question of interest here is can spun silk yarn be used to make functional silk strings?
How strong is the yarn - its breaking strength etc. Only one way to find out!

A macro image of the yarn shows its two strand twisted construction as well as its (less than ideal) 'hairiness'- not a problem for weavers of silk fabric but what about silk string makers?
A test string was quickly made by simply twisting 16 yarns together, 175 turns under a load of 240 grams (as for the experimental Hadd strings). The 175 turns was found to be the maximum degree of twist before the string began to lose its regular uniform cylindrical shape.
The string - still under load - was then coated with Gum Arabic to bind the silk fibres into a stable uniform string.
Some 'hairiness' remained so the string was then polished using a very fine abrasive (industrial grade 'rotten stone' ) on a cotton rag followed by another coating of Gum Arabic . This treatment resulted in quite a smooth and uniform string measuring 0.54 mm in diameter. (see attached macro image).

Due to its multi strand construction, the string is quite flexible compared to one made from simply twisted reeled silk filament - and the yarn, conveniently wound on a cone, is easier than reeled silk filament to handle as well.

A section of the sample string was then subjected to destructive testing - failing under a load of about 6 Kg - more than adequate.
The remaining length of string will now be tested on a lute to see how it performs.

If the spun silk does not perform well as a simply twisted string material it may well have a practical application for over spinning on a core of reeled silk filament to make.
More to follow later as time permits

Edited to correct the length of spun silk on the 200 gram cone - 13,000 yards (about 12 kilometers) not 1,200 yards as advertised.

In order to easily handle the spun silk on cone, a cone 'handler' is easily and quickly made (about an hour's work in total). The tapered cone holder has been made from a piece of scrap pine - rough cut to the required taper on a band saw - and then finished by hand planing to fit the cone. (I could not be bothered to turn the tapered holder on a wood lathe). The cone rotates on a simple bearing (just a wood screw) adjusted so that there is no 'binding' of the cone when rotated. This unsophisticated design is perfectly OK for low speed operation.
The spun silk will be fed directly from the cone to the string winder for making each string bundle.

Although spun silk yarn shows some promise as a material for making silk strings it is likely not an historical alternative.

So, returning for now to historical considerations.

The Mersenne-Taylor Law for vibrating strings that are thin, cylindrical and otherwise uniform mono filament in structure gives the fundamental frequency of vibration as a function of string length, tension and density.

Using the 'Old Oud' project as an example where string length is 56 cm, density 1.3 gm/cc and tension 3.5 kg, the maximum pitch of the treble Hadd will be about 370 Hertz (@A440) if frequent string breakage is to be avoided.
With the strings of a five course oud tuned a fourth apart this enables us to calculate string diameters according to Mersenne-Taylor and Kanz al-Tuhaf for comparison.
For those interested, the attached calculation sheets give the fine detail.

To summarise the results, there is a significant difference in string diameter between those given by Kanz al-Tuhaf and Mersenne-Taylor. For example, the Bamm string according to Kanz al-Tuhaf works out at 0.87 mm diameter whereas according to Mersenne-Taylor the diameter should be
1.4 mm.
This discrepancy can be explained in part due to the fact that silk strings are not mono filament but are made from a number of threads twisted together and consolidated with a glue binder. Twisting the threads together increases the string diameter - the greater degree of twist the greater the increase in string diameter.

The Kanz al-Tuhaf does not give details as to how the silk threads are twisted together to make a string. The most basic form of construction is one where a bundle of silk threads is simply twisted together. However, the preliminary tests to make simply twisted 16 thread (Hadd) strings and 64 thread (Bamm) strings reported in earlier posts confirm that the diameter increase resulting from simply twisting a bundle of silk threads to make a uniform string is about +20% maximum - sufficient to make the second (Zir) string but not the remaining Mathna, Mathlath and Bamm strings.
So, if the Hadd and Zir strings may be a simply twisted, what might be the alternative construction for the remaining strings?
More to follow.

Twisting thread bundles according to the thread count given by the Kanz al-Tuhaf does not result in the string diameters required by the Mersenne-Taylor law in the case of the Mathna, Mathlath and Bamm strings. So, is the information given by the Kanz al-Tuhaf incorrect? Can silk strings made from the thread bundle counts given by the KAT be made to work satisfactorily according to the Mersenne-Taylor law? The answer is very likely - yes !

The attached calculation sheets gives the fine detail.
To summarise, the KAT string bundles when twisted to the maximum amount - i.e. assuming a modest 15% increase in diameter from the untwisted string bundle diameter (to be further confirmed experimentally) - give string diameters of Mathna 0.71 mm, Mathlath 0.87 mm and Bamm 1.00 mm. Compare this to the required diameters given by the M-T law of 0.79 mm, 1.05 mm, and 1.40 mm respectively.

Solution (A) is to reduce string tension from treble to bass (string density remaining constant at 1.3 gm/cc).
This results in reducing string tension, according to the Mersenne-Taylor relationship, of Mathna
2.9 Kg, Mathlath 2.4 Kg and Bamm 1.8 Kg.

Solution B is to increase string density assuming constant tension. So, from the Mersenne -Taylor relationship, for the Mathna string a density of 1.6 gm/cc is required (23% increase) and for the Mathlath 1.9 gm/cc (46% increase and Bamm 2.6 gm/cc (100% increase).
From previous discussions posted about 'weighting' of silk with Tannin and metallic salts, these density increases should be well within the practical limits.

Of course, a combination of the (A) and (B) possibilities may be combined to produce any desired result within practical limits.

Next to examine a third possibility - that of adding additional material to a string by over spinning or wrapping a string with silk thread

The modern metal over spun oud bass strings, familiar to everyone, have cores of nylon filaments wound with wire to provide extra string mass without loss of string flexibility. Prior to the general availability of nylon in the 1950's, silk filament was used for the string core. A set of oud strings that I purchased in Cairo as late as the early 60's was comprised of gut trebles and wire over spun on a silk core basses.

The metal over spun string type first became available in the late 17th C. so were not used on ouds before this time (and likely not until much later).
However, the ancient Chinese made over spun bass strings with a silk core with a silk filament wrapping so it is possible that this type of string may also have been used on early ouds.

Wire wound over spun strings are made by feeding a soft metal wire onto the string core as it rotates at high speed under tension. An early string winding apparatus is shown in L'Encyclopedie' (1767) by Denis Diderot among the engravings illustrating 'Lutherie' (Planche XIII). The string is placed under tension with a weight and rotated at high speed with a hand cranked wheel that drives a hook at one end - the other end being connected to freely rotating swivel. The operator feeds the wire by hand onto the string (like a spring) as the core rotates.
This is essentially the same method used to make modern over spun strings except that the string core is mounted under tension between two synchronised hooks rotating together in unison (using an electric motor drive). The more "sophisticated" machines for mass producing strings are computer controlled with a wire feed carriage that automatically traverses the length of a string at the correct feed rate.

Mimmo Peruffo of Aquila Strings shows how it is done by hand in this video of a visit to the Aquila factory. Go to:

http://www.aquilacorde.com

Click on the FAQ button > Our Gut String Making > Video.
The demonstration is about half way through the video. It looks so easy!

However, this was not the apparatus used by the ancient Chinese silk string makers. More to follow.