A weave plan describes which yarn passes above the other at each crossing. The drawing-in draft assigns the warp ends to shafts, while the lifting plan tells those shafts what to do for each pick. A correct design needs all three to agree.
To check that agreement before yarn reaches the loom, trace each end through its assigned shaft, apply the lift for that pick, and compare the result with the intended crossing. This simple check can expose a wrong shaft number that an attractive fabric preview hides.
The examples below use an original eight-end, four-pick repeat to explain a conventional shaft loom, with one shaft assignment per end. They are not machine settings or proof that a finished fabric will meet its intended performance.
Read the cloth plan before the machine plan
An end is a warp yarn running along the fabric, and a pick is a weft insertion across it. In the usual point-paper view, columns represent ends and rows represent picks. A marked crossing commonly means that the warp lies above the weft on the stated face.
CottonWorks' preparation guide describes this design grid and its relationship to the drawing-in and chain plans. The three records answer different questions: the design shows the intended interlacement, the draft assigns the ends, and the chain or lifting plan controls the shafts over successive picks.
A drafting grid must be interpreted as a set of assignments: its marks connect ends to shafts rather than describing crossings with weft. Likewise, a lifting-plan column represents a shaft rather than an individual warp end, a distinction that can disappear when all three records use similar squares.
The distinction also appears in ISO 7211-1:1984, which covers presentation of weave diagrams and drafting, denting and lifting plans. ISO lists that edition as current following confirmation in 2023, and its public abstract also includes the arrangement of different yarns. The worked method here is an independent teaching example, not a claim to reproduce the full standard.
Write the convention beside the numbers
For this article, 1 means warp above weft on the chosen face, and 0 means warp below weft. All shaft instructions describe the state required during weft insertion: a shaft marked 1 is raised, whereas a shaft marked 0 is not raised in this simplified model.
Ends are numbered E1 to E8 from left to right in the chosen design view, while picks P1 to P4 follow the order of insertion. Shaft numbers S1 to S4 are labels, so they do not establish which physical shaft sits nearest the operator on an actual loom.
When transferring a design, keep that last distinction visible by checking the loom's shaft numbering, the software's viewing direction and the first pick in the file. A draft that is internally correct can still be loaded in the wrong orientation.
The WeavePoint complex-drafting tutorial states its own marked-square convention and display choices, including the possible change in marks for sinking-shed use. That is a useful reminder to declare the convention rather than assume that every file uses yours.
If another system uses 1 to mean a lowered shaft, convert the command meaning before sending it to the machine. Changing the column heading alone is insufficient, because the intended warp-over-weft crossing must remain the same after conversion.
An unfilled input is different from a deliberate zero. A zero is a known down state in this example, whereas a missing value is incomplete information. Resolve it before checking the plan; silently replacing missing data with zero can create a plausible but wrong cloth.
Build a repeat from the paths of the ends
Our repeat contains eight ends and four picks, so it has 32 crossings. To make each four-pick path easy to read, the following numerical table lists one end per row, although the usual point-paper diagram would put the ends in columns.
End | Assigned shaft | P1 | P2 | P3 | P4 |
|---|---|---|---|---|---|
E1 | S1 | 1 | 0 | 0 | 1 |
E2 | S2 | 1 | 1 | 0 | 0 |
E3 | S3 | 0 | 1 | 1 | 0 |
E4 | S4 | 0 | 0 | 1 | 1 |
E5 | S3 | 0 | 1 | 1 | 0 |
E6 | S2 | 1 | 1 | 0 | 0 |
E7 | S1 | 1 | 0 | 0 | 1 |
E8 | S4 | 0 | 0 | 1 | 1 |
Reading E5 from P1 to P4 shows that it must pass below, above, above and below the four picks. E3 needs exactly the same sequence, so both can be controlled by S3 in this model.
The other matching pairs are E1 with E7, E2 with E6, and E4 with E8. There are four distinct end paths, even though the repeat contains eight ends. The resulting drawing-in sequence is:
S1, S2, S3, S4, S3, S2, S1, S4
This is not a straight draw through four shafts repeated twice. The reversal in the middle is deliberate, and replacing it with a familiar sequence would change the crossings.
Why this repeat needs at least four shafts
Because all ends on one shaft move together, it cannot give two ends different states on even one pick. The CBSE Woven Textiles teacher manual, in its section on drafts and peg plans, explains this grouping principle. It also allows matching ends to use separate shafts for practical reasons.
For our model, the four different paths require at least four independently controlled shafts. Because the number of picks alone does not determine that minimum, count the different complete end paths over the full repeat.
This is a logical minimum, not a recommendation to use the fewest possible shafts in production. A mill may distribute identical paths across extra shafts for an approved practical reason, provided each duplicate shaft receives the matching lift sequence. The reed, yarn density, equipment and setup still need review.
Before extending this rule beyond the stated shaft model to leno controls, layered structures or a jacquard harness, examine how those systems actually control the ends.
Derive the lifting plan from those groups
For each pick, look at the required state of one end from every shaft group, then confirm that all other ends in that group agree. P1 needs S1 and S2 raised, followed by S2 and S3 on P2; continue the same comparison through the full four-pick cycle.
Pick | Shafts raised | Shafts not raised |
|---|---|---|
P1 | S1, S2 | S3, S4 |
P2 | S2, S3 | S1, S4 |
P3 | S3, S4 | S1, S2 |
P4 | S4, S1 | S2, S3 |
After P4, the sequence returns to P1, and that return needs checking as part of the repeat just like the first three changes.
For example, E6 belongs to S2. On P2, S2 is raised, so E6 is above the weft. On P3, S2 is not raised, so E6 lies below it. Both results match the design table.
The plan describes the required shed selection without prescribing lift height, shaft timing, dwell, acceleration or loom speed. Since those settings depend on the machine and yarn system, a correct list of raised shafts does not justify a change in mechanical timing.
CottonWorks' weaving-basics guide explains how heddles on a shaft move their ends together and distinguishes shaft shedding from jacquard control. The important point for this example is the shared motion: a shaft cannot repair a mistaken assignment of one of its ends.
Reconstruct every crossing instead of checking only the picture
Give the draft and lift plan to a second person without the intended design table, and ask them to reconstruct the 32 crossings. This makes the check independent of the picture they expect to see.
The relationship can be written compactly:
Crossing at pick p and end e = lift state at pick p of the shaft assigned to end e
If D(e) is the shaft assigned to end e and L(p, s) is that shaft's state, then:
W(p, e) = L(p, D(e))
Here W is the reconstructed crossing, with the same 1/0 convention; the equation is an indexing rule, not a calculation of yarn force or fabric strength.
For E5 on P3, D(E5) is S3 and L(P3, S3) is 1. The crossing is therefore 1. For E5 on P4, the same shaft has state 0. Changing the pick changes the lift state; it does not change the end's threading.
A spreadsheet or a short program can repeat this lookup across every cell while checking for invalid shaft assignments. Every end needs one valid shaft number, and every pick needs a defined state for each used shaft. Report the location of each mismatch, not just a total pass or fail.
AdaCAD's documented drawdown operation similarly combines threading with a lift plan. Its documentation warns that, if a threading column has several marked cells, the operation uses only the one nearest the origin. A rendered output cannot replace input validation because a program may resolve an ambiguous input without establishing what the designer intended.
A threading error and a lifting error leave different clues
First suppose E5 is mistakenly assigned to S1 instead of S3. Its intended path is 0, 1, 1, 0, but S1 gives 1, 0, 0, 1. All four crossings for that end are wrong, producing errors that run down one end column in the usual grid.
Next restore the correct draft, then mistakenly leave S3 down on P2. Both E3 and E5 are affected, because both belong to S3. Two crossings on one pick change from 1 to 0, so the grid shows errors across that pick at the affected ends.
These logical signatures do not provide a complete fault diagnosis, since a broken end, missed pick or incorrect physical setup can add other patterns. Use the design comparison to decide which record to inspect first, then confirm the actual threading and machine state.
The four wrong cells in the first example are 12.5% of this 32-cell design record. That percentage measures disagreement between two small technical plans, rather than predicting a garment defect rate.
Check the join between repeats
A float is a run in which one yarn passes over successive yarns of the other set. Because a repeat boundary can divide that run on paper without breaking it in the fabric, check a tiled repeat instead of an isolated rectangle.
E1 has the path 1, 0, 0, 1. Its final 1 joins the first 1 of the next repeat. The warp therefore floats over two successive picks at that boundary. Looking only inside the printed row could miss that join.
For the weft on P2, the crossings from E1 to E8 are 0, 1, 1, 0, 1, 1, 0, 0. Under our convention, the weft is above wherever the entry is zero. The final two zeros join the first zero in the next width repeat, creating a weft float over three successive ends.
Checking all paths gives a maximum warp-over run of two picks and a maximum weft-over run of three ends for this repeat. Those counts of crossed yarns are neither measured free-float lengths in millimetres nor universal acceptance limits.
In this simple single-layer design, investigate an end that never changes state or a pick that leaves every end on the same side. Passing these elementary checks still does not prove that any arbitrary binary pattern will make stable cloth.
AdaCAD's tutorial on drafts and graphics makes this distinction clear: a compelling black-and-white pattern need not retain the same appearance or integrity when woven. A drawdown expresses crossings without fully modelling yarn size, movement or the finished fabric.
Do not confuse the reverse face with a mirrored plan
If the same identified crossing is viewed from the other face, warp above becomes warp below. With end and pick identities held fixed, its binary state changes from 1 to 0, or from 0 to 1. That is different from reversing the order of the end columns.
Since physically turning a sample can reverse its reading direction, mark the face, warp direction and starting corner before comparing it with the plan. Otherwise, a correct back-face view can be mistaken for a threading error.
Software may offer mirror, inversion and rotation commands together, although a mirror changes positions and an inversion changes which yarn is above. Check their effect on one identified crossing before applying them to the complete repeat. Keep that crossing's identity through the comparison, rather than choosing a new convenient corner after each change.
Keep denting separate from shaft assignment
Denting places the ends through spaces in the reed, controlling their grouping across the reed width. A shaft assignment controls their up-and-down movement. The two plans must agree on end identity, but they serve different purposes.
In our eight-end example, a two-ends-per-dent arrangement would put E1/E2 in the first dent, E3/E4 in the second, E5/E6 in the third and E7/E8 in the fourth. Although the first dent contains ends on two different shafts, passing through one reed dent does not make those ends move together.
Two ends per dent is not a universal rule, because the chosen reed and yarn must suit the intended construction. A denting sequence with different occupancy needs its own record, including deliberate empty dents, rather than being inferred from the lifting plan.
An original reed-width check
Assume an illustrative body warp has 1,440 ends, with two ends per dent in a reed specified as 12 dents per centimetre. This is an explicit unit basis, not an unspecified traditional reed count.
Body dents = 1,440 ÷ 2 = 720 dents
Body width in the reed = 720 ÷ 12 = 60 cm
Average end density in that reed area = 12 × 2 = 24 ends/cm
The body contains 1,440 ÷ 8 = 180 complete weave repeats, confirming the arithmetic alignment of the chosen body-end count with our eight-end repeat.
Now add 16 selvedge ends at each edge, also assumed to use two ends per dent. Each edge needs eight more dents. The complete warp has 1,472 ends in 736 dents, occupying about 61.33 cm in the reed.
The selvedge structure has not been designed by that calculation, so it still needs its own threading and lift instructions. The 60 cm body width also cannot predict finished fabric width without measuring the changes during weaving and finishing under the specified conditions.
Even equal average end density can conceal different denting. A repeating occupancy of 2, 1, 2, 3 ends uses eight ends in four dents, just like 2, 2, 2, 2. Despite the equal average, the local grouping differs, and each arrangement's suitability depends on the yarn, reed and intended fabric.
Keep yarn colour and structure in separate records
A structural 1 does not mean a dark yarn, and a structural 0 does not mean a light yarn. The interlacement remains present when both yarn systems have the same colour, and a later colour change need not alter their shaft assignments.
Record the warp and weft material sequences beside the structural plan, giving each colour or yarn type an unambiguous code tied to the actual material specification. A digital colour swatch is not a yarn-lot identity.
Suppose our eight-end structure uses a three-end warp colour sequence. The two sequences return together after 24 ends, the least common multiple of eight and three. If the four-pick structure uses a six-pick weft colour sequence, they return together after 12 picks.
A 24-end by 12-pick display therefore includes a complete combined cycle. This is a safe repeat for checking both records together; visual symmetry may make part of the image appear to repeat sooner. Keep the actual sequence lengths as the basis of the check.
There are sixteen warp-above cells in the 32-cell structural repeat. That 50% share is not proof that half the fabric's visible area will show warp colour. Yarn diameter, spacing, flattening, crimp and finishing affect what the viewer sees, while a square grid assigns equal space to crossings that may occupy different areas in cloth.
ApparelDiary's yarn-count guide helps keep linear-density descriptions consistent when specifying the yarns. The same weave plan can give different fabrics when those yarns or their spacing change.
Decide whether an existing draft can make a new design
Reusing a threaded warp can save preparation work, but first test whether its groups fit the new design. In our draft, E1 and E7 share S1. Any replacement lifting plan must give those ends the same path, so changing it alone cannot make them differ on P2 in a proposed new design.
Check every shared-shaft group against the proposed repeat, since a conflict requires a different assignment or a different design. Do not hide it by selecting one end as the representative and ignoring the others.
Shaft names themselves are arbitrary. If S1 and S4 are renamed everywhere, both in the draft and in the lift-plan labels, the reconstructed cloth stays the same. Renaming them in only one record changes the result, making this a useful test after a software conversion or a loom-numbering change.
Before reusing the design, check its yarn sequence, reed plan and actual end count, because a matching interlacement repeat cannot prove compatibility with a new stripe layout or selvedge.
The starting pick needs similar care. A cyclic shift of an endlessly repeated structural sequence may describe the same uncoloured cloth, but it changes the reference point. If a weft-colour sequence stays fixed while the lift sequence shifts, compare them together before deciding whether the changed relationship affects the intended cloth.
After a stop, the record should identify the next insertion to be made, not simply the last screen row displayed. Use the loom's approved restart procedure to establish that position. Because a duplicated pick and an omitted pick are different events, preserve the actual sequence for comparison with the sample. That record should reflect what happened rather than what the operator hoped to resume.
Approve the physical sample against the plan
Before weaving, retain both the intended crossing record and the exported file, then reconstruct the latter using the stated origin and shaft convention. A screenshot of the design before export cannot prove that the machine file contains the same instructions.
At the loom, confirm the end order, shaft assignment and denting through the approved setup procedure. Use authorised machine checks; a design review does not justify reaching into a moving mechanism. Record the first-pick position so a later stop or restart can be compared with the correct part of the cycle.
Inspect a woven sample on the declared face and in both yarn directions. Compare several repeats and their boundaries, check the selvedges separately, then assess dimensions and appearance in the required finished state. An exact interlacement match is necessary for this design, but it does not establish strength, comfort or suitability for use.
ApparelDiary's fabric roll acceptance checklist can support the later receiving record. Keep that production assessment distinct from the 32-cell design check, since it records the cloth actually made rather than proving agreement between the technical instructions.
When a sample differs, locate the first mismatch by end and pick, then follow that end to its shaft and that pick to its lift command. This gives the team a specific path from the visible crossing back to the instruction that should have produced it.
Discussion0
Loading the discussion…