A staggered shed costs whoever the threading raises
Worth reading first: The quietest shafts belong at the back · The shed is an extension · The back shaft works hardest.
The quietest shafts belong at the back found that the threading which sorts a harness by flex count, busiest in front, gives the least worst end, and that a satin stripe on a plain ground is already sorted by its straight draw and cannot be improved. Its two plain shafts change sides eight times a repeat and its eight satin shafts twice, a ratio of four, while a ten-shaft harness’s strains differ by only 1.76 from front to back. No threading can absorb a flex spread larger than the harness’s own strain spread. That account then asked whether a loom setting could do what a threading could not: whether a staggered shed, with different shafts lifted different amounts, could flatten the stripe’s product.
It cannot, and the reason is short. What a stagger can do turns out to be a different question with a more useful answer, about which ends pay for a stagger that a weaver wants for another reason, and the threading decides it.
The clean shed is already the least any end can have
The shed is an extension fixed the shape of a shed from one condition. The shuttle crosses at the reed, so what the loom must guarantee is a clear opening there, 30 millimetres on the ordinary broad loom these essays use, with the reed 90 millimetres from the fell. A warp end runs straight from the fell to its heddle, so its height at the reed is its heddle’s lift scaled by the reed’s fraction of the way. Requiring the same opening at the reed from every shaft forces each shaft’s lift to be proportional to its distance from the fell. That is the clean shed: every end at the same angle at the fell, and the back shafts lifting furthest.
The strain on an end is the extension of its two straight runs, fell to heddle and heddle to back rest, over the warp’s length, and to leading order it is the square of the shed’s tangent times the shaft’s distance from the fell over twice what lies behind it. Every part of that grows with the lift. So an end lifted less than the clean shed would have a smaller strain and a smaller opening at the reed, and an end lifted more would have both larger.
That settles the question as it was asked. A stagger that lifted the back shafts less, to relieve them, would close their ends’ share of the shed below 30 millimetres at the reed, and the shuttle would strike them. The clean shed is the pointwise floor: it gives every end the least strain an open shed allows, so any change to the lifts leaves every end where it was or raises it. A loom setting cannot lower the busy plain ends’ strain, and those are the ends whose product is the worst in the stripe.
It can raise the quiet ends’ strain, and in principle that narrows the spread across the warp. But a narrower spread bought by working the quiet ends harder is not what the question was after. The worst end is exactly as worn as it was.
What a stagger is for, and what it does
A staggered shed is set for a reason that has nothing to do with strain. In a clean shed all the raised ends lie in one sheet and all the lowered ends in another, so as the shed closes and reopens, every end brushes its neighbours over their whole length. On a dense warp, or a hairy one, those neighbours catch: hairs from adjacent ends interlock, and the shed does not open cleanly. A hair layer is a balance, not a stock, and a warp in the loom has its hairs constantly raised by the reed and the heddles. A stagger sets alternate shafts at two heights, so that neighbouring ends travel in two separated sheets and do not cross in one plane.
The geometry follows from the floor. One group of shafts, the inner one, keeps the 30 millimetres the shuttle needs. The other is set to open wider, its top sheet higher and its bottom sheet lower, each by a stated height at the reed, . The inner group’s ends keep their clean strain and the outer group’s ends open by , so their strain rises by about the square of : 28 per cent at a 2 millimetre stagger, 96 per cent at 6.
So a stagger is a charge, levied on half the shafts, and what it costs the loom depends entirely on which half. The threading decides whose ends stand on the raised positions. That is the question a stagger actually asks, and it has the same arithmetic as the sort, with one more choice in it.
The straight draw puts a plain shaft on the raised level
Thread the satin stripe straight, plain shafts at positions one and two and satin at three to ten, and stagger it with alternate positions raised. Whichever group is raised, one of the two plain shafts is in it, because the two plain shafts are neighbours.
At a 6 millimetre stagger the straight draw’s worst end does 71.84 against 39.51 clean, 82 per cent more. The better of its two raised groups puts the front plain shaft on the raised level: that shaft has the least strain in the harness, so raising it costs the least, but its ends change sides eight times a repeat and carry the full factor on top. At 2 millimetres the same threading pays 20 per cent and at 10 it pays 157. The cost grows with the stagger as fast as the square in the strain does, because a busy end is paying it.
Now thread the plain shafts at positions one and three, and raise the even positions. Both plain shafts stand on the lower level and keep their clean strain; the raised positions, two, four, six, eight and ten, all carry satin. The worst end is the plain end at position three, which is one shaft further back than the straight draw’s second plain shaft, and that costs 7 per cent: 42.28 against 39.51.
Then the stagger stops costing anything. From 1.5 millimetres to 9.25, the split threading’s worst end stays at exactly 42.28. The raised satin ends do pay, and pay heavily, but they began at a quarter of the plain ends’ work and have a factor of about 2.6 to spend before the worst satin end at the back of the harness overtakes a plain one. Below 1.5 millimetres the straight draw is cheaper still, because a stagger that small costs its raised plain shaft less than one position’s worth of strain; above 9.25, the back satin shaft becomes the worst end and the cost climbs again.
Where the work goes along the harness
The profile shows the trade directly. Clean and straight, the work runs 36.8 and 39.5 on the two plain shafts and climbs gently from 10.6 to 16.2 along the satin. Under a 6 millimetre stagger threaded straight, the zigzag of raised and unraised positions appears on the satin, and the front plain shaft, raised, jumps to 71.8. Threaded for the stagger, the plain ends sit at 36.8 and 42.3 on the lower level, the raised satin ends at 19 to 32, and the unraised satin ends where they were.
Nothing about the cloth changes in either case. As a lifting plan says nothing without a threading, the lifting plan is rewritten for the new threading and the cloth off the loom is the same; the stagger is a setting of the shafts’ heights; the only things that move are which heddles are on which shaft frame and which frames are set high. Where the heddles go found the heddles’ own travel favouring a sorted harness almost everywhere, and a split threading moves one plain shaft back by a single position, which changes that travel by little.
Ninety placements, and the straight draw among them
The sort was a theorem with an exhaustive walk behind it. The staggered version has no tidy theorem, because the raised group is a second choice that interacts with the first, so it was walked directly. A placement is the pair of positions the two plain shafts take, 45 of them, and either alternate group can be raised, so a stripe under one stagger has 90 placements.
At 6 millimetres they run from 42.28 to 126.61, and the straight draw is no longer the best threading: twenty placements beat it. Every one of those twenty keeps both plain shafts on the lower level, and they differ only in how far back the second plain shaft has gone. The worst placements put a plain shaft on a raised position near the back, where the largest strain in the harness meets the stagger’s factor and the plain shaft’s eight changes all at once.
The rule that comes out of the walk is cruder than a sort and does the same job. Under a two-level stagger, thread the busy shafts on the lower level, as far forward as alternate positions allow. For a stripe with its busy weave on two shafts, that means positions one and three, and the price is one position’s worth of strain on one shaft.
A damask check has too many busy shafts
The damask check that a figure is harder on its warp introduced, an eight-end satin figure on an eight-end sateen ground, needs sixteen shafts: twelve whose ends change sides ten times a repeat and four that change six times. Sorted, its worst end did 90.83 clean, the busy shafts filling positions one to twelve. A two-level stagger on sixteen shafts has eight positions on each level, and the damask check has twelve busy shafts to place.
At least four busy shafts must stand on the raised level, and there is no plateau. The best threading rises from 90.83 to 101.40 at a 2 millimetre stagger and to 112.97 at 4, 24 per cent, with every millimetre adding more. Threaded straight it rises faster again, to 137.36 at 4 millimetres. The 3,640 placements of twelve busy shafts among sixteen positions, each with either group raised, were walked at each stagger, so these are the best the harness allows rather than a rule’s guess.
The best placement does something the stripe did not need to. The four busy shafts that must be raised go to the front, positions one, three, five and seven, where the strain they multiply is smallest, and the remaining eight busy shafts take every lower position to the back of the harness. The worst end is then a busy end on position sixteen, the back shaft, unraised.
At a 4 millimetre stagger, the best a damask check can do is 112.97, which is exactly what its straight draw did with no stagger at all. The coincidence is not one. In both cases the worst end is a busy shaft on the back position at its clean strain. The sort pulled that shaft forward and saved a fifth; the stagger, by halving the places a busy shaft can stand unraised, pushes it back.
How many busy shafts a stagger tolerates
The difference between the stripe and the damask is a count, and the count can be varied directly. Take a ten-shaft stripe whose busy shafts flex eight times and quiet ones twice, give it between one and seven busy shafts, and thread each as well as the stagger allows.
Up to five busy shafts, half the harness, the cost settles on a plateau. One busy shaft pays nothing at all until the back satin end overtakes it near 8 millimetres. Two pay the 7 per cent already found, and three, four and five pay a larger fixed step, 14 to 27 per cent, because spreading them onto alternate positions takes the last of them further back. Past half the harness some busy shaft must be raised, and the cost climbs with the stagger without limit: at 12 millimetres a six-busy stripe’s worst end does 2.30 times its clean value and a seven-busy one 2.49.
So whether a cloth can be staggered cheaply is decided by whether its busy shafts fit on one level, which is a fact about the draft before it is a fact about the loom. A plain ground under a satin figure, a twill border on a sateen body, any cloth whose hard-working weave takes a minority of the harness can carry a stagger for a fixed price. A figured cloth whose ground takes most of the shafts cannot.
How the numbers were reached
The loom is the one the earlier essays use: fell to first shaft 300 millimetres, sixteen between shafts, fell to back rest 1,200, reed 90 from the fell, 30 millimetres open at the reed. An end’s strain is the exact extension of its two straight runs at its shaft’s lift, over the warp’s length. A stagger gives one group of alternate positions an opening of at the reed and leaves the other at 30, so the shuttle’s clearance is kept. An end’s work is its flex count times its strain, in thousandths, as before.
The satin stripe’s flex counts, eight and two, and the damask check’s, ten and six, are those of the earlier essays. Every placement of the busier class among the positions, with either group raised, was enumerated at each stagger in steps of half a millimetre: 90 for the stripe and 3,640 for the damask check. The plateau’s ends were found in quarter-millimetre steps.
Required of the walk and not shown: that no placement under any stagger do better than the clean shed’s best; that the stripe’s split threading cost the same at 4 millimetres as at 8; that the damask check’s best under a 4 millimetre stagger lie above its clean best; and that a negative stagger, which would close the shed at the reed, be refused.
Where the arithmetic stops
The work is a ranking, not a fatigue law. As in what the shed costs in newtons, the product counts how often an end is strained and how far, and says nothing about how a yarn’s life falls with either. A raised satin end at three-quarters of a plain end’s work may break at a different rate from the plain end, and the plateau above is a plateau in the ranking.
What the stagger buys is not priced. The reason to stagger, ends that separate rather than cling, is real and nothing here measures it. A stagger of 2 millimetres may part a hairy warp as well as one of 8, or it may not; the arithmetic only says what each costs under a given threading.
Two levels only. Some looms stagger in three or four levels. The rule generalises in the obvious way, fewer positions on the lowest level, which tightens the count of busy shafts that fit, but those cases were not walked.
The strain is the shed’s alone. A background tension, the beat-up and the let-off add to every end alike, and a warp breaks in a tail no tensile test reaches: the breaks a loom sees come from the weakest places in a yarn, not from an average one. A stagger that raises the strain on a group of ends raises the rate at which their weak places meet a load they cannot carry, and how much is a question for that account.
Who worked out which part
Staggered sheds, split sheds and multi-level shedding are long-standing loom practice for dense and hairy warps, and the clean shed’s rule, lifts in proportion to distance, is a loom-setter’s commonplace. The flex-times-strain product and the sorted threading are the earlier essays’ own.
What is added here is the floor argument, that a clean shed already gives every end the least strain an open shed allows, so no stagger can level a stripe without working its quiet ends harder; the reading of a stagger as a charge on half the harness whose payer the threading chooses; the exhaustive walks, 90 placements for the stripe and 3,640 for the damask check, with the split threading’s fixed 7 per cent from 1.5 to 9.25 millimetres; and the count, half the harness, that decides whether a stagger can be carried at a fixed price.
Still open: how much stagger a hairy warp needs
The whole of this prices a stagger of stated height and says nothing about the height a warp needs. That height is a question about the hairs. Two ends in neighbouring heddles separate cleanly if no hair from one reaches the other across the gap the stagger makes, so the stagger a warp needs is set by how far its hairs stand out: hairiness goes as the root of the count, and a yarn’s distribution of hair lengths has already been worked out for exactly this kind of question, as a hair layer is a balance, not a stock showed for wear.
If the gap a stagger needs at the reed is about the length of a yarn’s longer hairs, a millimetre or two, then the stripe’s plateau is wide enough for any warp and a split threading carries the stagger for 7 per cent. If it needs much more, the plateau runs out and the satin ends overtake. The reading would come from the hair model’s own tail set against the shed’s geometry at the reed, and it would turn this account’s free variable into a number.
Shares its objects with
Essays naming at least two of the same things, that neither author linked.
- A harness reads a warp's tail best threaded at both ends — both name harness, heddle, shed, threading
- Strong faults move a break log's floor into the harness — both name harness, heddle, shed, threading
- A second wrong digit is silent only by cancelling the first — both name census, harness, threading
- A stripe is a partition of the warp — both name census, heddle, threading
- The harness has a depth — both name harness, shaft, shed
- The shortest notation has the longest mistakes — both name census, harness, threading
Named objects
A flat tag is an object no other essay names yet.