r^'^.^!^;:^^ ■'■■■ ;':.^^"-':r''\- Its 1669 .S35 I Copy 1 In 3@(c m^m 53{D)[cr: Classification of Raw Siik Bv Mechanical i es ts WAT?RRK i □J[C Hiaji w^m Classification of Raw Silk By Mechanical Tests Giving Rules Governing the Operation of the Gauge Reel and Cohesion Machine as Developed by Warren P. Seem Copyright 10 W by SILK PUBLISHING COMPANY Preface THE purpose of this pamphlet is to phicc at the disposal of those adopting the Gauge Reel and Cohesion Machine for the measurement of Evenness, Cleanness and Cohesion, a set of rules and tables with which to determine the relative value of each quality, reduce same to a common value and show its application in the classification of raw silk. I desire to acknowledge the courtesy of the Klots Throwing Co., in the development of the Gauges and the Schwarzenbach Huber Co., in the development of the Cohesion Machine, at whose expense and direction these tests were brought to the present stage of perfection in the interest of and for the benefit of the silk trade of the United States. The author has been dulv authorized to perfect the machines and place them on the market at his own expense. This pamphlet is furnished free to those purchasing a Cohesion Machine or a set of Gauges. As numerous calls have been received for these tables and rules, the author is obliged to ask a remittance of $1.00 per copy from those not ordering machines so as to cover this initial expense and provide funds for the standardization of tests. Warren P. Seem. Altoona, Penna., May, 1919. '" " (e)CI.A515937 4 M 13 1919 ^c'^^ ^ Classification of Raw Silk Evenness Evenness of the cloth is in a direct relation to the num- ber of fine and coarse threads found in a fixed yardage as I propose to show by the graphic description following: - - - »...!...■■■■£■■■■■..■■>■■■■■!■.■■■■■■■■■■■■■:■■■■ FIG. 1. The sketch shows the fine and coarse threads woven as the filling, single, in the proportion of 5 fine to 1 coarse, with 3 regular size threads between, to present a gooa illustration. As the purpose of the sketch is to show the relation of fine and coarse threads to the evenness of the cloth, the frequency in which the regular threads occur is immaterial and is therefore not given in the same propor- tion as they appear in the test as that would make the sketch too large. Each square represents the cross sec- tion of a thread, the fine thread to represent 10 deniers. the coarse 20 deniers. the regular size 15 deniers. It will be observed that where the coarse sizes appear the unevenness is worse but we cannot view the increase in the thickness of the cloth from the thin part made by the fine thread but must view it from its average thickness made by the regular size thread; a fine thread makes the cloth thin and weak, whilst a coarse thread makes the cloth thick, and when twisted is dull in shade, or causes streakedness. A cloth to be even must lack both the ex- treme fine and coarse threads; viewing them both from the average thickness of the cloth, the one is as serious as the other, therefore it will be observed that the less fine and coarse threads there are the evener the cloth and the more fine and coarse appear the more uneven the cloth. In throwing Organzine when two threads are doubled together of the proportion 1 to V/,, even with the most even doubling, the coarse size will wind around the fine and give the thread a corkscrewed appearance; the greater the difference the more marked will be the corkscrew. The very fine threads, however, present another condition and that is they overstretch and do not contract the same as the regular size and make a bad corkscrew, therefore again we have the same relative condition and the evenness of the Organzine is in proportion to the number of fine threads found in the raw thread. The fine, in addition to making the threads thinner and corkscrewed, also makes the thread 33 per cent, weaker, that is the fine thread gen- erally breaks 33 per cent, lower than the combined strength of both threads of average size. The coarse thread twists up a duller shade, causes streakedness and gets stuck and breaks in the reed when using a fine reed. Can the evenness be determined by the ecart or spring of sizing skeins? No, as the short fine and coarse even up. Can the relative evenness be determined by the weight total of the sizing skeins above and below the average and their aggregate weight be compared with that of the same number of skeins of average weight or anj' other mathematical computation from the sizing? No, as the more short uneven lengths there are the more they even up as the following illustration shows. A lot of Shinshiu No. 1 showed on sixty sizings. skeins of 225 metres a range of from ten to sixteen deniers or a spring of six deniers. A Gauge test showed on 300,000 yards: 135 Very Fine threads under 6 deniers 315 Fine threads under 10 deniers 210 Coarse threads over 20 deniers One of the most uneven silks ever tested. Here is the situation: when the fine and coarse threads are very few, say twenty to every 300.000 yards, then they come only once on the average in every 15.000 yards and generally get into one sizing and give a false relative value and when they are many they get so close together that they average up with the coarse and again give a false relative value. In valuing evenness we must remember first that it is only the extremes that count in working results and the quality of cloth. A very fine thread which always breaks either in throwing or weaving is just as serious from a working point of view if only one or two inches long as when 100 inches long. 100 very fine are 100 times worse than one very fine. Take the fine ranging from six to ten deniers on a 13/15 denier thread and the coarse over twenty deniers are the ones that pass through and get into the cloth and cause the unevenness that becomes noticeable and lowers the quality of the fabric. In valuing evenness we need consider only the threads one-third under and over the average size and our test for evenness must be one that will show^ this truly in the easiest, simplest and quickest way. The fine and coarse threads average between five inches and twenty- five yards. One fine or coarse streak five inches long would not be as bad as if it was twenty-five yards long, but 180 fine streaks five inches long or the equivalent of the twenty-five yards length would be decidedly worse that the one spot twenty-five yards long, so also would 600 streaks in fifteen yards of cloth be much worse than thirty streaks of the same average length. The point I desire to make is that unevenness is in proportion to the number of very fine, fine and coarse threads regardless of their length and that a test for evenness must show these three conditions regardless of their length, which cannot be done with sizing skeins no matter what length they are reeled. A number of Japanese Inspectors were approached on the uniformity idea shortly after the several articles were published and they said that they could meet the uniformity idea with No. 1 stock and according to the results we have had recently they appeared to have accomplished it very well, indeed, much to our regret. Is then the practice of judging the evenness by the spring faulty and this custom that has been in vogue for years without any merit? No, when the fine and coarse threads are few they come so far apart that rarely more than one fine or coarse gets into one sizing skein, and they either increase or decrease the weight of same as the case may be; then the spring is a true indicator of even- ness. Experienced inspectors never used it as a definite method of judging evenness, only as a sign or an assist- ance to the inspection. I have found it holds true only in inspecting XX and XXX silk and when the stock is under these two gi-ades then the fine and coarse threads come frequently enough to even up, even on 225 meter sizings. and no dependence can be placed on same. Should the same tables for evenness be used on coarse sizes as on fine? No, as it is easier to reel a coarse size evener than a fine size as the following example shows: We will assume that a cocoon fibre average 21/2 deniers; a 11/13 denier would then be the product of 5 cocoons and the running out of one fibre would affect the size 2,5 -^ 12.5 or 20 per cent. On a 25 denier it would affect the size 2.5 -^ 25 or but 10 per cent; a 7 denier 2.5 -h 7 or 36 per cent. A tabulation of the following tests shows that as the sizes get coarser the evenness becomes better. 8 Lots 16/18 Very fine Pine 63 Coarse 11 23 Lots 18/20 Very fine Fine 64 Coarse 18 14 Lots 24/26 Very fine Pine 30 Coarse 4 5 Lots 28/30 Very fine Fine 12 Coarse 1 Upon the basis of this information and a summary of a great number of lots on the various sizes I have divided the evenness tables into 3 divisions, thus: 11 to 16 deniers 17 to 22 deniers 23 to 28 deniers and over 7 points = 1% 5 points = 1% 3 points ^ 1% Very fine to represent 7 deniers and under on all sizes, but they are to be increased as follows: 11 to 16 deniers x 4 17 to 22 deniers x 5 23 deniers and over x 6 The evenness table to be as follows: 11/16 17/22 23/28 = 100% 828 = 40% = 100% 368 = 40% — 100% 276 = 40% Method of Determining Evenness by Gauge During the development of the Gauges, before I had the diameter of raw silk as given by Rosenzweig, it became necessary to know at what opening of the Gauges different sizes of silk should be tested; this I determined by taking 10 threads from 10 different bobbins and drawing about 20 yards rapidly through the Gauge by hand and shitting the thread to various openings of it until I obtained a distinct draw on the thread. I then averaged up this result and tested the silk at the average size as determined by method named. During the translation of the French 1904 edition of Serivalor in 1910, I discovered the diameter of raw silk as given by Rosenzweig, and set the Gauges accordingly and was surprised to find that we agreed within one denier. During these tests I discovered that Cantons gave the same draw as hard natured Japans on a size about 1'/, deniers coarser, and that Japan with a strawy hand gave the same draw at from 1 to 2 deniers coarser. The filling required on this nature thread was found to be from 10 to 12 per cent, less, which confirmed the result obtained by the Gauges. I now use on Cantons a 16 Gauge for a 14 silk and strawy Japan a 15 Gauge for a 14 silk. On the Japan the difference in results are so close that there is no need for being over critical on this point, as there is always some doubt as to the average size of the raw tested. As the thread flattens out somewhat in going through the Gauges, extreme accuracy is not required as to the Gauge number. It can readily be understood how coarse sizes choke up in the Gauges and break down the thread, but as to the fine and very fine, it generally requires a practical demonstration to convince testers that they actually do catch the fine threads. A fine thread is the result of allow- ing one or more cocoons to run out; then when two or more cocoon fibres are added, by a cast on the running thread, the diameter is increased beyond the average size of the raw which catches and breaks down the thread; as fine threads are due to carelessness of the reeler, that same carelessness also causes many other defects, all of which catch and show up the fine thread. The fine end is not, however, a direct catch of the Gauges, as other defects, but must be looked for on the reel or take-up side ot Gauge. Originally it was planned to get the fine threads by putting an adjustable tension on the paying-off bobbin, regulating it according to the size of the thread, but this was found unnecessary, as the per cent, of fine threads not recorded in a test was made under 1 per cent.; besides as the tester does not tie out all of the fine but just enough to get a thread strong enough to tie up to, there was a question of doubt even to the 1 per cent, error shown. Originally we counted the very fine as 50 per cent, under average size and the fine as 33 per cent, under average size. Skilled silk workers judged this relation of evenness by feel and sight, but in teaching learners a definite method became necessary, which was accomplished by tieing a knot on thread thus; cent, between double the result of a 50 centimetre thread single; the thread was then wrapped once around the post and this result on even silk was exactly double that of the 50 centimetre length single; a test conducted on very uneven silk showed but a variation of 3 per cent.; as the 3 per cent, variation was undoubtedly due to the variation in the silk thread tested, I adopted as the maximum breaking point of very fine threads double, 8 x 7 deniers to 56 gi-ams; the minimum fine as per following table. All very fine threads then are based on a 7 denier double or 7 X 4 X 2 = 56 Grams. 10/13 den. 11/13 den, U/U den. 13/15 den. 14/16 den. 1,S/17 den. 16/18 den. lS/20 den. 20/23 den. : 22/24 den. 24/26 den. 26/28 den. 2S/30 den. : : Avg. 11 ■ : Avg. 12 ■ : Avg. 13 - : Avg. 14 : Avg. 15 ■ = Avg. 16 : Avg. 17 ■ : Avg. 19 - : Avg. 21 ■ : Avg. 23 - : Avg. 25 - : Avg. 27 - : Avg. 29 - 33% = 7.36 den x4x2 — .58.85 call 59 Grams. 33% = 8 den. x4 X 2 zz 64 call fr4 Grams. 33% =r 8.71 den. x4x 2 zz 69.5S call 70 Grams. 33% =: 9.38 den x4 x2 =: 75.0-1 call 75 Grams. 33% := 10 den. x4x 2 := 80 call 80 Grams. 337o = 10.72 den x4x 2 zzi 85.76 call 86 Grams. 33% =z 11.39 den. x4x 2 zzz 91.12 call 91 Grams. 33% z= 12.73 den. x4x2 = 101.84 call 102 Grams. 33% zz 14 den. X 4x 2 1= 112 call 112 Grams. 33% = 15.41 den. X 4 X 2 i^ 123.2S call 124 Grams. 33% z= 16.75 den. x4 X 2 =; 134 call 134 Grams. 33% z= 18 den. x4x2 := 144 call 144 Grams. 337o = 19.43 den. x4x2 = 155.44 call 155 Grams. A scale was then rigged up as per following sketch, which also shows method of using same: FIG. 2. and when the knot passed through at Gauge number at which the silk was tested it was considered a fine; this, however, still left the very fine subject to the judgment of the operator and on coarse sizes different operators showed different results. Comparisons between working results and the Gauge tests showed that it was necessary to fix a standard size for the very fine and let it represent a size that breaks out in throwing and in weaving in the gum, single, on all sizes of raw and that a positive method of determin- ing the very fine was necessary. This condition was met by calling the very fine 7 deniers and under and determin- ing same by their breaking strength. A silk thread break- ing at 4 times its average diameter, expressed, in grams, is considered 100 per cent, strong, therefore I set the breaking strength of a very fine at 4 x 7 or 28 grams on all sizes, the maximum breaking point of fine to be 33 per cent, less than the average size tested. As many of the fine threads are only a few inches long it is impos- sible to use the regular length of 50 centimetres, nor would It be practical as it requires too much time to fasten each thread to the two posts of Serimeter each time a fine thread is found. To save time and adapt ourselves to the length of thread available, a series of breaking tests were con- ducted by passing the thread over a i/j-inch post of the Serimeter and drawing down on both ends of thread until they broke. This method showed a difference o£ 13 per The uncleanness of a silk thread is represented by the following defects: Slugs Nibs Bad Knots Waste Split Threads Bad Thows Corkscrews Hairiness Cocoon defects Reeling defects Slugs and nibs are made by the worm and we call them cocoon defects. A nib is a small slug; when they are about the size of two raw knots, or are oblong, then call them slugs; if very large call them large slugs. The purpose is to distinguish between the large and small defects and reduce their relative value accordingly. The rest are reeling defects and made by the reeling girl. The following prints show the different shapes and sizes ot defects named above except corkscrews and hairiness. Corkscrews are so well know that no illustration appears necessary. Rosenzweig calls them "rognose." Hairiness consists of small loops about 1/64 inch long, and because of their great number look like short hairs standing up on the skeins when looking across the face of same. This hairiness shows in the cloth wnen the loops appear about 30,000 to 45,000 on 300,000 yards of thread. They do not catch in the Gauges, if they did it would be very impractical to count them, even when under 30,000 on 300,000 yards of thread. Instead of including them under Cleanness I penalize the quality number 1 per cent, for 40,000 to 60,000 and 5 per cent, when over 200,000 per 300,000 yards. These may be counted on one section of the reel after a layer of silk has been reeled and increased to 300,000 yard basis. The Gauges have been condemned by critics because they do not catch these minute defects, also because they say that when the thread runs fine then the smaller defects are not removed as thoroughly as when the thread runs coarse and that a perfect instrument to measure Cleanness of the raw silk thread must adjust itself to the variation of the silk thread. Instead of being a fault I find it a distinct advantage. The very worst silk has over a million defects per x yards (300,000). To attempt to count these would be impracticable, besides what benefit would be gained by the knowledge if many thousand of them do not show in the cloth. In counting defects on mirrors no two inspectors would give the same results as the one having the keenest sight would find more defects than the one of duller vision. The thread is run through the Gauges at the average size of the thread tested which, for example, we will take as 14 deniers; let us now suppose that the thread becomes 10 deniers; theoretically it appears that it will not clean the thread as thoroughly as it did when the thread was running through at 14 deniers, but we must not forget that the decrease in diameter is not in proportion to the denier count or as 14:10 but in proportion to the area of a cross section which in turn is in proportion to the square of the diameter. 10 deniers = 0.00183 inches 14 deniers = 0.00217 inches or as we see a 10 denier is but 15 per cent, less in diameter than a 14 denier silk. Let us look at it another way: if we cut a thread in two and look at a cross section we see that as it increases in size it spreads out in all directions thus and when it goes through the Gauges it exposes but two sides to the Gauge and in addition to that is the flatten- ing effect on the thread as it passes through at high speed. It is quite evident that what we want to know is the relative number of objectionable defects that are on the thread; their size governs this exclusively; therefore, no matter on what size thread they appear the only thing we want to know is, are they large enough to be classed as objectionable. How are we to determine this? By the sense of vision the result varies with the inspectors. A more definite means is therefore necessary, which has been found in the Gauges. These are set to a standard size according to the diameter of raw silk as found by Rosenz- weig and given in Serivalor, which are as follows: Deniers Microns Inches Deniers Microns Inches 8 42 0.00165 32 84 0.00328 9 44 0.00174 33 85 0.00333 10 47 0.00183 34 86 0.00338 11 49 0.00192 35 87 0.00343 12 51 0.00201 36 89 0.00350 13 53 0.00209 37 90 0.00355 14 55 0.00217 38 91 0.00359 15 57 0.00225 39 92 0.00362 16 59 0.00232 40 93 0.00366 17 61 0.00239 41 94 0.00370 18 63 0.00246 42 95 0.00374 19 64 0.00253 43 96 0.00378 20 66 0.00259 44 98 0.00386 21 68 0.00267 45 99 0.00390 22 70 0.00272 46 100 0.00394 23 71 0.00278 47 101 0.00401 24 73 0.00284 48 102 0.00402 25 74 0.00290 49 103 0.00405 26 76 0.00296 50 105 0.00414 27 77 0.00301 51 106 0.00418 28 79 0.00307 52 107 0.00422 29 80 0.00312 53 108 0.00425 30 81 0.00317 54 109 0.00429 31 83 0.00323 One Micron is .001 millimeter or 0.0000394 Inches. One Millimeter equal 0.03937 inches. The Gauge appears very simple, but here is one con- dition you must not overlook, and that is that the blades of the cleaner must be y., inch thick so the long soft slugs do not wriggle through the cleaners; the second difficulty is that as the difference between a 13 denier and 14 denier silk is only .00008 inch, it takes an extremely accurate cleaner and that it is impossible to set it to each size with an ordinary feeler blade and that feeler blades cannot be manufactured so fine. This problem has, how- ever, been solved by the author with the instrument known as the Silk Gauge, which is 2 inches wide, 6y„ inches long, and has a range of from 8 to 30 deniers andis ground to an accuracy of one ten-thousandth part of an inch. Experi- ments show that it is unnecessary with the Gauge to put any tension on the thread, but as all fine threads are accompanied by a defect larger than the average size, they break down and are recorded. This method is to provide a mechanical inspection test to substitute the inspection of mirrors and overcome the confu- sion due to difference in vision and judgment of various in- spectors in different parts of the world. Nothing is left to the judgment of the operator, it is wholly automatic in its operations. The Amount of Silk Required to Get Constant Results, Duplicate Tests and Represent the Lot Twenty skeins selected from different books of the bale have been found to represent the bale; when less than that is taken the results are reliable if the rest of the bale is like the sample, but to get the average condition and detect variation, 20 skeins are necessary to a bale. Of course you can fool this by putting in a number of books of very bad silk, but for this the tester must be on guard and when one skein runs very much worse than the others then another set of 20 skeins must be taken from the books that have been omitted in the first selection. To represent the lot two tests out of five, three out of ten and four out of twenty bales are necessary under present conditions to get the average of a lot. When the run of the bales are known then one test will be sufficient. The fewer the defects the more silk is required to get constant results as the defects are further apart and consequently a sufficient amount of thread must be used to permit these to average up: the closer the defects the less amount of thread is required. The result of several thousand tests and comparative working results show that when the total number of defects besides raw knots, fine and coarse threads equal 150, then the test can be stopped and the result increased to 300,000 yards by multi- plying by the following rule: 20 bobbins 500 yds. ea, total 10,000 yds.x30 =300,000 yds. 20 bobbins 1000 yds. ea. total 20,000 yds.xl5 =300,000 yds. 20 bobbins 2000 yds. ea. total 40,000 yds.x 7yo=300,000 yds. 20 bobbins 3000 yds. ea. total 60,000 yds.x 5 '=300,000 yds. Care must be taken that each test represents the 20 skeins. As the Gauge Reel is only equipped with 10 ends change bobbins when test is half finished. Gauges The Gauges are made of a high grade tool steel, grad- uated, tempered, and then ground absolutely true. The fineness of the Gauges is such that the sprains of the metal must be considered and time allowed for this action to subside. (I have had a poor set warp .004 inch out of true during a cold spell where the proper metal was not used.) The first Gauges were made with blades % inch wide, but it was found that there was difficulty in setting them, as they sprung in setting, and when thus set they opened up on the machine. To overcome this they were made with blades % inch wide and later 1 inch wide, which solved the trouble. Method of Setting Gauges According to Rosenzweig, the diameter of raw silk of 12 deniers is .00201 inch and 27 deniers is .00301 inch; the nearest we can get to .002 inch is 12 deniers and .003 inch is 27 deniers, which are the setting points. It is impossible to measure the diameter of a silk thread with a Micrometer Caliper, as the thread yields or flattens out more or less under the pressure of the Micrometer Caliper. On account of this flattening out and the great variation on silk of the same size, no correction can be determined upon. The only way we can correctly set the Gauges so that the slot in the Gauges corresponds to the outside diameter of the silk thread is by feel and to make this as near uniform as possible it was found necessary to use weights. It also was found that by using a feeler blade y„ inch wide the same feel could not be had at the lowest point of the Gauge as at the high and also that the measurement was so sensitive that in getting the feel, the Gauges, when made with but i/, inch blades, were opened up. To avoid these errors the feeler blades were then made but 3/32 inch wide and weighted as fol- lows: .002 weight 14 grams .003 weight 17 grams The ordinary feeler blades bought on the open market are not accurate enough for this work and they must be selected with a Micrometer Caliper measuring by ten- thousandth part of an inch. (No. 75 Micrometer Caliper made by Brown & Sharpe Manufacturing Company is used by the inventor of the Gauges.) The Gauges are shipped properly set, but if by a mishap they require resetting, tighten up the rear bolt dead tight, then tighten up the front bolt until the .002 feeler blade holds fast at 12 deniers and the .003 feeler blade at 27 deniers. Draw the feeler up and down in slot until it holds its own weight. First clean out the slot very care- fully, getting rid of every particle of dust — remember you are working on a very fine measurement. A good grade of tissue paper I find the best for the purpose. Be careful it does not break oft' and stick fast in the Gauges. Graduation The graduations are an arbitrary measurement and based on 15 equal subdivisions between 12 and 27 deniers. The Gauges were made 6i/o inches long so as to make the graduations to cover from 8 to 30 deniers. Operating Gauge Test Speed of Reel The speed of the takeup of the Reel Fly is to be 250 yards per minute. Care of Gauges The Gauges must be kept absolutely clean and free from rust or scum. Oil daily with a good quality of spindle or clock oil, making sure that the oil has run between the two faces of the Gauge blades. Do not permit the Gauges to become choked up with waste, as it is possible to force open the Gauges slightly if waste is wedged in tightly. When the Gauges get choked up with silk thread then use a 1.5 feeler blade to push or pick it out. Put the blade in back of the place you desire to clean out and move it towards the front or open part of Gauge. Do not use a bent feeler blade that scratches the Gauge. Method of Making Tests First — Wind from unsoaked silk about 3,500 yards from each of 20 skeins on 20 bobbins, 10 from the under side of skein and 10 from the regular side. Take every other bobbin and threaden up on the Gauge Reel. Shift the Gauges so they all run at the average size of thread. The operator must constantly watch the threads and stop the reel as soon as a break occurs so as to get the yardage uniform. A black mirror should be placed in back of the threads so as to see the threads plainly. Cohesion Cohesion is the term applied by Rosenzweig in Seri- valor to the force that causes the two cocoon filaments to stick together as one compact thread. In physics, cohesion is considered as the mutual attraction of the particles of a solid for one another, and is measured by the amount of force which must be applied in order to overcome it. The term cohesion is generally applied to the mutual attraction of particles of the same substance; adhesion to that of different substances. We thus see then that the term is well applied. The silken fibre (sericin and fibroin) consists of two filaments (bavelle, brin) which when issued by the worm through its spinneret are laid side by side and agglutinated together as one cocoon fibre. The force of this agglutina- tion we call Cohesion. The threads that are agglutinated very tightly are dense and firm and resist opening a long time; these we call a very good Cohesion; those that are loosely agglutinated and open very readily we call very poor Cohesion. The f.ict that the cocoon fibres consist of two filaments enables us to measure the Cohesion of the cocoon fibre, but as we use only the reeled thread we are not interested commercially in the Cohesion of the cocoon fibre except as it relates to that of the reeled thread and its relative value in classifying the physical qualities of raw silk. My experiments on raw silk thread reeled with a long, medium and short croisure or twist indicate that Cohesion i.s more dependent on the cohesiveness of the sericin than the length of twist in reeling, yet a good twist is essential and must be long enough to knead the thread well together so that it will be well agglutinated and yield a thread with a good Cohesion. Further observations show first, the importance of hav- ing cocoons uniform in quality second, that Lustre is co- related to Cohesion; third, that 90 per cent, of tests show Tenacity and Cohesion practically alike in their relative value. Investigations are under way to show reasons why 10 per cent, of tests show a variation in the relative value of Cohesion and Tenacity of about 10 per cent.; whether it is due to an en-or in the method of making the test, relative value of Cohesion and Tenacity tables, or an actual variation in the relative value of these two named qualities. All reelers agree that to produce a thread with a high Cohesion requires a high grade seed, close attention in sericulture, careful and skilled reeling. It will be observed that since the requirements of a thread with a high Cohe- sion enhances its value, if in the classification of raw silk we base its physical qualities on Cohesion, then we have a basis that presents a relation true to its real cost to produce and is therefore worthy of your serious con- sideration. In weaving single thread in the gum the cocoon fibres constituting the thread must be well agglutinated; if one or more fibres are loosely gummed together then when the threads rub against each other as the shed opens and closes or the harness or reed rubs the thread, then these loosely agglutinated threads open, split off one or more fibres and frequently break. In skein dye we find that silk with a very poor Cohesion has a tendency to split up in dyeing and cause Lousiness, also that when the sericin is boiled off it produces a skein very wooly or exceedingly loopy in appearance. In spinning Organzine with a thread of low Cohesion sometimes one or more cocoon fibres split off, run a band on bobbin, cause excessive breaks, labor cost and waste. Measurement of Cohesion Even though in the loom we find that it is the friction of the harness and reed on the thread and that of the thread upon itself as the shed opens and closes that causes the thread to open and split off, yet extensive experiments show that we cannot measure Cohesion by friction because of the rough character of some silk threads due, I am told, to reeling the thread from cocoons that have been softened in water containing sand and limestone salts. Also there are threads which are so tightly agglutinated that they rub through before opening; still others are so smooth and silky that they resist opening by friction a long time and give very untrue relative results when compared with actual working qualities. The method of testing resorted to commonly is to sep- arate the fibres of the thread between the thumb and fore- finger, noting such results as this crude procedure may make manifest, the personal equation rendering the thumb and forefinger test, at best, no more than a means for ar- riving at approximate results, which, although better than nothing, are not capable of tabulation and comparison when secured by the efforts of different operators, or even by the efforts of the same operator, working at different times or under different conditions. The sericin or gum or raw silk not only possesses various degrees of adhesive- ness, hut, also, frequently contains minute particles of foreign matter, such as sand or limestone, which roughens the sericin to such an extent that conclusive results as to cohesion cannot be attained by a mere friction test. In view of the foregoing, it is the object of this invention to provide a simple but efficient machine whereby cohesion tests on silk thread may be carried out readily, and whereby the results of the tests will be of such a con- clusive and uniform natm'e that they may be compared, thereby producing a standard for cohesion tests. Experiments covering several years show that Cohesion can be measured by rolling the thread under pressure. I then followed the rule of physicists and measured its relative order, fixed an arbitrary scale of relative values. The number of strokes required to roll open the thread constitutes the unit of Cohesion. The test is made on 200 threads, 10 threads from eacli of 20 skeins. See Fig. 1 for Cohesion machine. Reeled threads requiring but 286 stroltes to open showed an open condition of the thread under a magnification of about 150 diameters lilve Fig. 3. 509 strolves see Pig. 4 827 strokes see Fig. 5 1612 strokes see Fig. 6 It appears that when the tlireads have a Cohesion over 800 strokes then they are so tightly agglutinated that no voids or openings appear. i-k;s. j. 4, 5, 0. Rules for Operating Cohesion Machine Speed The speed is to be 108 strokes ( full stroke back and forth) per minute. Five strokes more or less will not affect the results. Oiling The roller must be kept oiled at its V shaped bearing. Care must be taken not to get oil on roll. It should spin around freely when moved with the finger. In the worm gear casing use a mixture of 50-50 cylinder and machine oil; keep it about % full. Oil slide and reciprocating motion about once a week. Adjusting Roll The roll must be level and roll evenly on the mirror; care must be taken in adjusting it so as not to bind the roller and retard its movement. This can be tested by moving the roller with the finger and seeing if it spins around freely. The roller must bear uniformly on card the full length of stroke and full width of card. This can be tested by running the machine on a blank card and observing whether the polished place on card, when rolled. Is of the same brightness. Use paper liners under card carrier lever. Care of Roller It is absolutely necessary that the roller be kept free from dust, scum, rust, oil, perspiration and free moisture. Before using the machine clean off roller with a piece of chalk and wipe off with a dry rag. Black Cards or Mirrors Use a good quality of hard black cardboard and avoid using a grade that varies very much in thickness or in its hardness. Soft cardboard retards the opening ofi threads about 2 to 3 per cent. Use grade like sample shipped with machine. Preparation of Mirrors Use an ordinary sizing reel upon which reel, with a spacing of about 60 threads to the inch, 10 threads from each of 5 skeins or a total of 50 threads to a card. Place the mirror under the threads, then paste a gummed label on each end of mirror, gluing down the threads firmly onto the mirror. Make 4 cards for each test or 200 threads all told. Operation of Cohesion Test Fasten one mirror firmly on each Cohesion machine, start same and then place threads on roller. If the thread opens up with two or three scratches with the nail of thumb then the card should be examined after the first hundred strokes and every 50 strokes thereafter until all of the threads are opened the full length of the stroke. If it takes 6 to 9 scratches to open the thread, then the machines may be run about 1000 strokes before the cards need be examined. When the threads once start to open then the card should be examined every 50 strokes. To examine the threads swing back mirror carrier on rest, remove mirror, bend same so as to slacken the threads, push in a Spatula and bend up against threads and see if all threads are opened the full length of stroke. If not, replace the mirror and continue the test for 50 or more strokes until all the threads are opened the full length of stroke. The number of strokes required to open all the threads fully is the unit of Cohesion for that mirror. The unit of Cohesion for the test is the average of the four cards representing 200 threads. Note that we use the maximum of each mirror, not the average, but as we use 4 cards the average is obtained as the unit of Cohesion. To take the average of each card requires more time to make a test and a different table must be used. The relative value would not be any different. Air Conditions The test should be conducted at a relative humidity of from 65 to 70 per cent, at about 70 to 75 deg. F. No test should be made when humidity is over 75 per cent, or under 50 per cent. The lagging effect of a silk thread is such that if the mirrors are put in a small box with 65 to 75 per cent, humidity for one hour then they may be tested under any reasonable temperature without any ap- preciable difference. Starting New Machine In starting a new machine rub roller with chalk and run it on a mirror without threads about 4 hours so as to remove any gi-it or corrosion that might have accumulated on roller in transit. Cohesion Table Based on the maximum average of 4 cards of 50 threads each on unsoaked silk. Per Strokes Cent. Classification Application 2200 100 2100 99 Extra Good Cohesion Very Good for Weav- 2050 98 ing in gum single 2000 97 Extra Good Lustre Very Good Organ 1950 96 1900 95 1850 94 Very Good Cohesion Good foi Weaving in gum single 1800 93 Very Good Lustre Very Good Organ 1750 92 1700 91 1650 90 Good Cohesion Good for Weaving 1600 89 gum single 1550 88 Good Lustre Good Tram 1500 87 1450 86 1400 85 Fair Cohesion Fair Organ 1350 84 1300 83 Fair Lustre Good Tram 1250 82 1200 81 1150 80 Fair Cohesion Fair Organ 1100 79 1050 78 Fair Lustre Good Tram 1000 77 950 76 900 75 Only Fair Cohesion Poor Organ 850 74 800 73 Only Fair Lustre Fair Tram has a ten- 750 72 dency to become 700 71 wooly 650 70 Poor Cohesion Very Poor for Organ 600 69 550 68 Poor Lustre Boil off carefully 500 67 450 66 400 65 Very Poor Cohesion Has a tendency to get 350 64 lousy and woolly 300 63 250 62 Very Poor Lustre Very Poor Tram. 200 61 150 60 Note: Cohesion over 2200 strokes usually is very high in lustre and hand. Winding Test A true winding test can only be made by counting the breaks on the whole skein from start to finish. As Japan skeins at 1G7 thread speed run from four to five liours, it talves too long a time to malce a test on tlie wtiole skein and a shorter test becomes necessary. After working several methods for a number of years I found that any short method, that would be true to the breaks on the whole skein within a maximum variation of fifteen per cent, would be close enough for a winding test. I then made a series of 20 tests on Japan, single and double skeins. Italians, Chinas and Cantons, and from this series found the fol- lowing method came within the 15 per cent, ma.ximum limit; the series did not vary more than 5 per cent. Air Condition — Temperature 70 to 75 per cent, at 70 to 75 per cent. Relative Humidity. First — Select 20 skeins from one bale. We must make our selection from original packages, that is one bale. Whilst most of us buy in five and ten bale lots, the bales are not always alike. A bale, however, generally represents the product of one filature which is not always true of a five or ten bale lot. I have found 20 skeins necessary to represent the average conditions of a bale and give results equal to tests made on 60 skeins. I found 10 skein tests varying too much. „ , ^ Kub (jums Gums or reel markings must be rubbed out before wind- ing. On hard gums they can be daubea with a warm emulsion which softens the gums so they can be thoroughly removed. oi • ■ t-« i • J>keining or Uandering Skeins up 10 skeins with under side up, the other 10 regular. „ , Speed Rated when test is half completed. On unsoaked silk with Gums 120 Thread (Yards per minute). On unsoaked silk without Gums and soaked silk ISO Thread (Yards per minute). Swifts Twelve stick. Pin Hub unweighted. Method of Making Test Make a starting run of fifteen minutes, don't count breaks, then wind 300,iiOO yards counting breaks. At 120 thread speed. Run 125 minutes. At 180 thread speed. Run 84 minutes. The sum of all the breaks on the 300,000 yards equals wind- ing count. (Foi- waste and cost tables add eight to count for first end tied np wliicli is the equivalent uf tlic twenty first ends tied up on tlie ^^(ill.COO yards wound.) The breaks the first half hour are very excessive at times. Creak tests made only on the first hour winding sometimes vary as much as 100 per cent. When making a starting WINDING TEST SHEET FORM NO. I la freted Speed Test Started St'ijDped . 1 f-st Tunc ^'^^ X/ .X-' .-^V /// Coarse .'/ / SL'MMARY ACniAL we.we TABUS Waste ,\, , :, Small Raw Knots >>, -s^ -^ --^ ,s -s, -*^ -.-v .7^ -V. >. -:^ -^^ .-^ ^ -.. -***. -x ■■-s •% -".^ '*> A •>, r^ ->^ V BaK -X -^ >, >, .^. Locips Split Ends V Corkscifw Bod Throw3 ^ [n I>>ul>t ; /;.'- U ^ if.^J' w Cfiunt 18/20 21/22 23/24 25/26 27/28 29/30 31/32 33/34 35/36 37/38 39/42 43/46 47/50 51/54 55/58 59/62 63/66 67/70 71/74 75/78 79/82 83/86 87/90 91/94 95/98 99/102 103/106 107/110 111/114 115/118 119/130 131/140 141/150 151/160 161/170 171/180 181/190 191/200 201/210 No. .Swifts Uini 114 110 105 100 94 90 86 82 78 75 72 66 62 58 54 52 50 47 44 42 40 39 37 36 35 33 32 31 30 29 28 27 26 25 24 23 22 20 19 inding Table Waste 13/15 Den Amer. Keel 0.30% 0.33 0.36 0.39 0.42 0.45 0.48 0.51 0.54 0.57 0.63 0.69 0.75 0.81 0.87 0.93 0.99 1.05 1.11 1.17 1.23 1.29 1.35 1.41 1.47 1.53 1.59 1.65 1.68 1.74 1.87 1.92 2.17 2.32 2.47 2.62 2.77 2.92 3.07 St-(."anton Reels Class Very well 1.80% 1.84 1.95 2.06 2.15 2.24 2.33 2 42 2.50 2.59 2.86 3.08 3.33 3.52 3.74 3.96 4.18 4.20 4.50 Very well Well Well Well Well Well Fair Fair Fair Fair Only fair Only fair Only fair Poor Poor Poor Poor Very poor Very poor Very poor Well for Canton Well for Canton Well for Canton Fair for Canton Fair for Canton Fair for Canton Fair for Canton Poor for Canton Poor for Canton Poor for Canton Poor for Canton Very poor for Canton 10 Evenness Table As tlie relative value of each defect varies, it is necessary to reduce them all to a common value, which is done in the following manner: Evenness. — Very fine on 11/16 denier multiply by 4 Very fine on 17/22 denier multiply by 5 Very fine on 23/30 denier multiply by 6 Fine. — AH sizes multiply by 1. Coarse. — All sizes multiply by 4. Average Yokohama Classification Three per cent, under Best Class 11/16 Den. 17/22 Den. 23/28 Den. % 100 9 4 3 99 Very 18 8 6 98 Good 27 12 9 97 36 16 12 96 45 20 15 95 54 24 18 94 63 28 21 93 Good 72 32 24 92 81 36 27 91 90 40 30 90 99 44 33 89 108 48 36 88 Fair 117 52 39 87 126 56 42 86 135 60 45 85 144 64 48 84 Only . 153 68 51 83 Fair 162 72 54 82 171 76 57 81 180 80 60 80 189 84 63 79 198 88 66 78 207 92 69 77 Poor 216 96 72 76 225 100 75 75 234 104 78 74 243 108 81 73 252 112 84 72 270 120 90 71 288 128 96 70 306 136 102 69 324 144 108 68 342 152 114 67 360 160 120 66 378 168 126 65 396 176 132 64 414 184 138 63 432 192 144 62 450 200 150 61 468 208 156 60 486 216 162 59 504 224 168 58 522 232 174 57 540 240 180 56 558 248 186 55 576 256 192 54 594 264 198 53 612 272 204 52 630 280 210 51 Exceedingly 648 288 216 50 Poor 666 296 222 49 684 304 228 48 702 312 234 47 720 320 240 46 738 328 246 45 756 336 252 44 774 344 258 43 792 352 264 42 810 360 270 41 828 368 276 40 Cleanness Table Cleanness. These defects are all penalized according to their size and seriousness. All sizes are alike but table of common values different. Penalties for Cleanness Waste X 1 Nibs X 1/20 Very large slugs x 1 Slugs X 1/2 Bad throws x 1 Very long knots x 1 Long knots x 1/2 Loops X 1/20 Very large loops x 1/20 Corkscrews x 1/20 The tables on evenness and cleanness are made up for common values. Average Yokohama Classification Three Per Cent. Under Best Class 11/16 Den. 17/22 Den. 23/28 Den. % 100 15 11 8 99 Very 30 22 16 98 97 Good 45 33 24 60 44 32 96 75 55 40 95 90 66 48 94 105 77 56 93 Good 120 88 64 92 135 99 72 91 150 110 80 90 165 121 88 89 180 132 96 88 Pair 195 143 104 87 210 154 112 86 225 165 120 85 240 176 128 84 255 187 136 83 Only 270 198 144 82 Fair 285 209 152 81 300 220 160 80 315 231 168 79 330 242 176 78 345 353 184 77 Poor 360 264 192 76 375 275 200 75 390 286 208 74 405 297 216 73 420 308 224 72 442 323 236 71 464 338 248 70 486 353 260 69 508 368 272 68 530 383 284 67 552 398 296 66 574 413 308 65 596 428 320 64 618 443 332 63 640 458 344 62 662 473 356 61 684 488 368 60 706 503 380 59 728 518 392 58 750 533 404 57 772 548 416 56 794 563 428 55 816 578 440 54 838 593 452 53 860 608 464 52 882 623 476 51 Exceedingly 904 638 488 50 Poor 926 653 500 49 948 668 512 48 970 683 524 47 992 698 536 46 1014 713 548 45 1036 728 560 44 1058 743 572 43 1080 758 584 42 1102 773 596 41 1124 788 608 40 Tenacity Test Relative Humidity should be maintained between sixty- five to seventy per cent, at a temperature of seventy to seventy-five degrees F. Select twenty skeins from different parts of a bale and reel 22y„ meters from each of the twenty skeins, making a full length sizing skein of 450 meters. Weight this care- fully in deniers and use it as the average size of thread. Cut this skein in two and draw out thirty threads from different parts of the skeins, making the Tenacity Test on a standard Serimetre in the usual manner. To find the relative value, apply strength table or multiply the aver- age size by four and divide this result into the average tenacity, multiplied by 100, shown by test, c. g., Size 14; Tenacity 56; thus: 56 X 100 14 X 4 = — 100% 56 As this test is made for the physical qualities of the thread, not the structural, we must avoid all threads having structural defects in this test. 11 STRENGTH TENACITY TABLE Class fo DENIEES 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 106 47 51 56 60 64 68 72 76 80 84 88 92 96 100 104 108 112 116 105 Very Good 104 103 102 101 100 44 48 52 56 60 64 68 72 76 80 84 88 92 96 100 104 108 112 99 98 S7 95 93 41 45 49 53 57 61 65 67 71 75 79 82 86 90 94 98 101 105 92 Good 91 89 87 86 39 42 45 49 53 57 61 63 67 70 74 77 81 84 88 91 94 98 8b 84 Pair 83 81 80 35 38 41 44 48 52 56 59 62 65 69 72 75 78 82 86 88 91 Only- Pair 7^ 77 76 74 73 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 71 70 Pnnr 69 67 66 29 32 34 37 40 43 46 49 52 55 58 60 63 67 69 71 74 77 65 Very Poor 64 63 61 60 27 30 32 35 37 40 42 45 47 50 53 55 58 60 62 65 67 70 12 Raw Silk Classification Altoona. pa. Feh. 28th, t9i 9 LOT NO. Sample mark Sample bale no. Sample STOCK Japan chop Sa.Tiple GRADE SIZE 14.85 ArPLAR4NCE COLOR LUSTRE HAND OR TOUCH ILMRli-NESS PENALTY SKEIN GUMS WHlTIi IVORY S CKEAtI VERY EVEN EVEN FAIRLY EVEN UNEVEN VERY UNEVEN VERY GOOD GOOD X FAIRLY GOOD I'AIR POOR VLKY I'ilim.. . .. ... VERY SILKY SILKY NERVY X FTRM STRAWY SIONGY VERY BAD percent. BAD 3 per cent. GOOD ST. AMERICAN IMITATION STRAIGHT CROSSING OTHERS SOFT HARD NARROW BROAD MEASURABLE QUALITIES BASED ON 300.000 YDS. MADE ON 60 . 000 YDS. ICW lO--.* IQte' -^b^ ItSS iQin Total AVERAGE EVENNESS CLEANNESS WINDING COUNT . 13 GAUGE TEST RAW KNOTS ( "> f8{ (9) (11) (7) (5) (in ) (501 (2501 VERY FINE BKNG, STRAIN UNO. 24 OR KINE. 33'r UNDER AVG SIZE ?, 2 4 1 p n f;i=- ■^' 55 COARSE. 33.C OVER ^4 WASTE 1 ], ^ \1 p VERY LARGE KNOTS 2 ,1 .1 1R 1 h LARGE KNOTS NIBS 1 h b ;■; 4 ii p.q T /.!-, X .., 7 VERY LARGE SLUGS ,\ 1 SLUGS 1 1 1 1 4 P.O 10 VERY LARGE LOOPS .\ '.„ LOOPS 14 10 B ij^ 9 lb 68 340 ^ .'-•o 1 V SPLIT ENDS BAD THROWS -^ ^ p 1 3 ]? An X 1 60 CORK SCREWS 11 H 15 17 13 l^) 97 48P X '..n P.C TOTAL DEFECTS 1125 TOTAL i^il, PER CENT. Cj ,1 TOTAL 138 PER CENT. QT TENACITY TEST GE,1MS COHESION TEST RAW COHESION TEST SOAKED 55 1775 RELATIVE VALUE °0 pe, „nt 93 RELATIVE VALUE per cent QUALITY SCALE BASED ON RELATIVE VALUES QUALITY SUMMARr RELATIVE VALUES STRENGTH 95 % COHESION 93 % AVERAGE 94 % EVENNESS 94 % CLEANNESS 91 % COMMON VALUE ALL QUALITIES 93 % PENALTY % QUALITY No 93 % REMARKS: fc GRADE ."f GRADE too 99 98 XXX 97 96 SI 80 Ben 79 No 1 78 77 95 94 XX 93 92 76 75 74 No. 1 73 91 90 89 X 8S 87 71 70 69 I lo 68 U 67 66 86 85 BcM No 1 8» to 83 Eilra 82 65 64 63 No. 2 62 61 6U SIGNATURE A.E. THROWING' DYEING OE WEAVING REPORTS 13 liAL'GE REEL FOR DETERMIXINC E\EX.\EbS A.\U CLEANLINESS OF SILK :ii i» COHESION-MACHINE Patented April 15, 1919 PLATE 3— SMALL RAW KNOTS. PLATE 4— BAD KNOTS. 14 (1, w 15 a z w o I H a. w H w < Cm o o ^J < 1-1 16