A thread can pass through a needle eye and still sew badly. The row may break at a folded edge or leave cut yarns in the cloth. It may look sound before washing and distorted afterward. A useful selection process checks the whole seam assembly, including how it is made and used.
Size charts help narrow the choices. They cannot describe every thread construction, fabric finish, layer stack and machine setup. The final choice needs a clear trial. Another person must be able to repeat its setup and apply the same decision rules.
This guide develops that trial from the garment requirement to the approval record. Its numerical cases are invented teaching examples. They show how to compare candidates and judge the evidence. A general size table cannot guarantee performance.
Define the operation before choosing a size
Start with the seam's job. A visible topstitch, an internal joining seam and a stretch opening do not ask the same things of the thread. Record the stitch and seam construction and the layers being joined. State how the seam must move in use.
Describe the difficult locations as well as the straight run. A pocket corner, folded hem or crossing seam changes the layers under the needle. A good row through two flat layers proves little about these thicker junctions. Include them in the trial.
The fabric description needs a lot reference and its current processing state. Identify whether the material has already been dyed, softened, coated or fused. If another finish is planned, include it in the trial route. A swatch from an earlier development stage may answer a different question.
American & Efird's thread-selection guidance lists sewability, seam performance and appearance as separate needs. Supply and price also matter. That is a useful starting point for the brief. A low thread price cannot settle whether the seam performs its required job.
Write the trial question in a form that can be answered. For example: “Which approved candidates can sew this pocket assembly? Which meet its damage, strength and care requirements?” “Find the best thread” leaves too much open.
Put thread size on a consistent basis
Record the supplier, product name, construction, size system, colour and lot. Keep needle thread, bobbin thread and any looper threads in separate fields. A change in one position should not silently become a change in all positions.
Coats' thread-numbering guide explains that tex expresses grams per 1,000 metres. A larger tex number therefore means more mass per unit length. Read a commercial ticket number with its product type. The same ticket number can mean different sizes across thread types.
Consider two hypothetical finished threads labelled 30 tex and 40 tex. At their nominal linear densities, 200 m would have these masses:
30 tex thread: 30 × 200 ÷ 1,000 = 6 g.
40 tex thread: 40 × 200 ÷ 1,000 = 8 g.
The second carries (40 − 30) ÷ 30 × 100 = 33.3% more mass per length. This does not establish a 33.3% increase in diameter, breaking force or seam strength. Construction, material and finish still matter.
These are nominal comparisons, not a laboratory test of two cones. If measured linear density is needed, use the agreed method and specimen state. The package's gross mass includes its support. Dividing that mass by an assumed thread length cannot prove the stated tex.
Do not compare a single component yarn with the finished sewing thread as though they were the same item. A supplier's product data may state the component count and number of plies. Ask which figure gives the finished thread's count. Use that basis for all candidates in the comparison.
ApparelDiary's yarn-count systems guide explains the wider unit relationships. For this trial, keep a single clear basis rather than converting every label from memory. Preserve the original label beside the common unit.
Read a supplier chart as a product-specific reference
A chart should name the thread construction and explain its needle column. A minimum recommended needle size is a starting limit. It is not a required size for every fabric that uses the thread.
The A&E apparel thread comparison chart, version 01.19, illustrates this point. Its 40 tex entries for Perma Core Ultimate and Perma Core list different minimum needle sizes: Nm 80 and Nm 90. The common tex value does not produce one universal minimum across those products.
The same chart identifies its strength figures as comparative averages, not minimum requirements. Do not copy one of those averages into a buyer's seam specification. The chart concerns named thread products. A thread result is not a finished-seam result.
Use the current data for the actual product when making a purchasing or trial decision. The dated chart above shows how to read the information. It does not confirm current supply or approve either product for your garment. Keep the supplier's relevant document and revision with the trial brief.
If a chart uses broad labels such as light or heavy fabric, add the details it lacks. A dense thin weave can be harder to enter than a loose thick knit. A laminated stack presents another task. Their names alone cannot establish the same needle choice.
Check the needle system before testing size
The needle must belong to a system accepted by the exact machine. Coats' needle guide distinguishes system, point and size. It also describes the eye and the protective thread route along the groove. Loose thread passing through the eye is only one small check. It does not show how the pair will work while sewing.
Keep the full needle code, including any special geometry or surface designation. Within an approved system, shortlist sizes with the supplier and machine guidance. Do not admit a look-alike needle from another system just because its size number matches.
Point shape is a separate choice. SCHMETZ's round-point guide distinguishes normal round points and several ball-point forms. The light ball point is designed to move textile yarns aside. That intended action does not remove the need to inspect the actual sewn material.
Start the size comparison with the same point family where possible. Keep other specified features the same too. If the point changes too, describe the result as a comparison of complete needle products. Do not attribute every difference to diameter alone.
Install and orient each candidate using the correct machine instructions. Keep adjustment work with qualified personnel and the machine safely stopped as required. If timing or alignment changes during the trial, record it. The comparison now includes that change.
Use the complete material stack
Prepare assemblies that reproduce the seam allowance, folds, backing, tape and any interlining. Include the actual direction of sewing relative to the fabric. Mark each junction. Later observations can then refer to that same feature.
Fabric mass per area is useful, but it does not fully describe that stack. A loose thickness reading by hand also differs from the stack under the presser foot. Avoid inventing a rule that a given GSM always requires one needle size.
Keep material sources balanced between candidates. Suppose one combination gets soft roll-edge pieces and another gets centre pieces. Fabric location then varies with the combination. Distribute comparable pieces across the trial and retain their identities.
Use fresh, unsewn paths for each candidate. Do not run several needles through the same holes. Later candidates would face a different starting condition. Keep seams opened for inspection apart from specimens reserved for strength testing.
Include any required stretch direction or layered junction in the sample plan. Tugging each specimen until it looks acceptable does not give them all the same treatment. State the conditioning and handling that will precede inspection.
Set the decision rules before seeing the samples
Separate the requirements into checks with clear outcomes. Check for thread breaks and cut fabric yarns. Assess appearance, strength and performance after care as separate outcomes. A candidate must meet each required condition. One good result should not cancel an unrelated failure.
Define what an inspector will count. A thread break is an event during sewing. A damaged assembly is a unit containing the specified damage. One assembly may contain several damaged sites, so its unit count is not the same as its defect count.
Agree the viewing method for appearance and damage. Record lighting, the sample state and any required magnification. Distinguish an open space between displaced yarns from confirmed cut yarns. If the observation is uncertain, mark it for further review. Do not count it as a pass.
The strength requirement also needs a method, direction and decision rule. Is the requirement a minimum individual value or a specified average? Which failures are valid results, and which indicate a testing problem? Resolve these questions with the buyer and laboratory before testing.
Supplier advice can help choose candidates. Approval still depends on the garment's stated requirements. Keep local trial limits clearly labelled. They should not later appear in a different product's file as an industry standard.
Build a small combination trial
For a woven assembly, suppose the shortlist has two thread products and two needle sizes. Product T30 is nominally 30 tex; product T40 is nominally 40 tex. The needle sizes are Nm 90 and Nm 100 within the same approved system and point family.
This is an invented case. Assume the supplier has checked all four combinations and considers each worth testing. These labels are not catalogue recommendations. Keep the bobbin product, stitch construction, fabric lot and other fixed conditions documented.
Combination | Needle-thread product | Needle size |
|---|---|---|
A | T30 | Nm 90 |
B | T30 | Nm 100 |
C | T40 | Nm 90 |
D | T40 | Nm 100 |
There are 2 × 2 = 4 combinations. This follows a full factorial comparison. Each chosen level of one factor is tried with each level of the other. It avoids judging T40 only with one needle while judging T30 only with the other.
Treat thread product as the factor. Products in one construction family can still differ in several ways. This trial does not isolate the effect of tex. The conclusion applies to the named products and setups tested.
Start with a stable machine. Record stitch length, feed arrangement, speed, thread routing and relevant tension checks. If a candidate needs a new tension setting to form a balanced seam, record that setting. You are then comparing usable setups with recorded settings. More than two physical variables may have changed.
Repeat the setup and control the order
Run two separate trial sessions in this example. In each session, sew four assemblies with each combination. Each assembly contains 0.5 m of new seam. The total is:
Assemblies per combination = 4 × 2 = 8.
New seam per combination = 8 × 0.5 = 4.0 m.
Across all four combinations, the plan contains 32 assemblies and 16.0 m of new seam. Record any repair sewing separately. Use the same length basis for every candidate. Do not add repair travel to just one candidate's denominator.
The NIST factorial example explains the value of repeated runs and a varied order. Where practical, randomise the four combinations' order within each sewing session. Record the actual sequence and any interruptions.
Eight assemblies are not eight independent machine setups. Four consecutive pieces share the setup used in that session. A repeat session can reveal setup variation. Even so, this small plan provides limited evidence. It does not establish a precise factory failure rate.
Check the retained reference before and after the session where appropriate. A damaged guide or changing machine condition can affect later samples. If the reference changes, find out why. Do not assign every difference to the candidate thread or needle.
Interpret the observations without averaging away a failure
Assume the completed teaching trial gives the following totals. Breaks refer to first-pass sewing of the stated new seam length. Damaged assemblies contain confirmed cut fabric yarns under the agreed inspection method.
Combination | Thread breaks | Damaged assemblies out of 8 | Breaks per new seam metre |
|---|---|---|---|
A | 2 | 0 | 0.50 |
B | 0 | 3 | 0.00 |
C | 4 | 0 | 1.00 |
D | 1 | 2 | 0.25 |
For A, the break measure is 2 ÷ 4.0 = 0.50 per new seam metre. For B, the damaged-assembly share is 3 ÷ 8 × 100 = 37.5%. The table deliberately keeps event counts, unit counts and length-based measures separate.
Suppose the agreed screen allows no damaged assemblies. It also permits no more than one thread break in 4.0 m. None of the four combinations meets both conditions. A and C exceed the break limit; B and D fail the damage condition.
Those are invented screening limits, not general garment acceptance levels. They show why zero breaks alone cannot identify the winning combination. B looks best on continuity but fails the other required check.
Do not turn these four rows into a universal rule about needle size. They suggest questions for this assembly. Examine the failed sites and check the setup records. Use that evidence to choose a controlled follow-up. A different point, thread product or seam design may require a new trial.
Eight assemblies with no damage provide limited evidence. They cannot prove the combination will never damage the cloth. Retain the sample size and conditions whenever the result is reported.
Look inside the pooled totals
Keep the session records beneath the summary. Suppose A had one break in each session. All four of C's breaks occurred in the first session. Each session contains four assemblies and 4 × 0.5 = 2.0 m of new seam per combination.
A therefore showed 1 ÷ 2.0 = 0.50 breaks per new seam metre in both sessions. C showed 4 ÷ 2.0 = 2.00 in the first and zero in the second. The pooled rate of 1.00 for C is correct, but it hides a large difference between the two runs.
That difference is a reason to inspect the records, not proof that the first operator caused it. Check sample identity and any package or threading changes. Review adjustments and the actual sewing order. An undocumented intervention between sessions limits what the comparison can establish.
If C needed a changed setup for its second run, keep those conditions distinct. The same thread and needle do not make the changed settings identical. Keep the runs separate in the record. The successful setup may merit a new controlled check. That success does not erase the earlier failed run.
Keep laboratory seam evidence distinct
A sewn row and a tensile specimen answer related but different questions. The actual ISO 13935-2:2026 scope describes a grab method for seam maximum force. The force acts perpendicular to a straight seam. It addresses woven fabrics and certain other eligible constructions, with stated exclusions. Ask the laboratory to confirm that the method fits the material and seam.
Naming a standard does not supply a buyer's pass value. Nor does it turn a hand pull into a standard test. The laboratory needs the applicable procedure, including sample preparation and conditioning. This guide does not replace those instructions.
Consider a separate hypothetical set of five valid force results: 176, 184, 190, 188 and 187 N. Their sum is 925 N and their mean is 925 ÷ 5 = 185 N.
Suppose the agreed rule requires every valid result to reach 180 N. The mean exceeds that figure, but the 176 N result is 4 N below it. Reporting only “average 185 N, passed” would apply the wrong rule. The example sets no standard specimen count. It does not make 180 N a universal requirement.
Record the failure mode with the force. Thread rupture, fabric rupture and opening caused by yarn movement need different interpretations. Only call the thread too weak if the evidence supports that cause.
Check the seam in its finished state
Coats' sewing-thread guidance considers both sewing behaviour and later performance. It identifies properties such as abrasion resistance, elasticity, colour fastness and shrinkage. Which properties matter most depends on the garment and its intended treatment.
Follow the specified pressing, finishing and care route for the trial samples. Keep an untreated comparison where the plan calls for one. A seam approved before garment washing may change afterward. Record both states to locate when the change appeared.
Inspect the same marked zones at each stage. Keep photographs or retained samples in the factory's record. Add a scale and clear sample identity. An overall photograph and a close-up show different levels of detail. Use comparable views when judging candidates.
For stretch assemblies, define the required movement and recovery check. For visible seams, include the agreed appearance assessment. One tensile-force result cannot establish all of those properties. Do not add a treatment the product will never receive just to create a dramatic comparison.
Resolve the failing mechanism before expanding the trial
Where a thread breaks, note its position and appearance. Check the approved thread path, guides, needle condition and the assembly at that location. A break at a thick junction does not by itself prove that the thread's nominal size caused it.
Likewise, inspect cut yarns and visible holes separately from loose stitches or puckering. The response should address the observed mechanism. A larger needle may reduce thread rubbing but create new fabric damage. A smaller needle may leave the original stitch-formation problem unresolved.
Change a limited set of documented factors for the follow-up. Suppose the fabric finish, point, thread and stitch density all change together. The new sample may be useful. It cannot show which single change caused the result. State whether the aim is diagnosis or validation of a complete new setup.
ApparelDiary's quality assurance and quality control guide places such evidence within the wider approval process. A trial informs that process. Production urgency cannot turn a failed requirement into a pass.
Confirm the candidate under the intended sewing conditions
After the small comparison identifies a suitable candidate, assess it over the agreed longer run. Include the actual difficult junctions and expected range of material variation. A few short pieces cannot show every effect that may emerge as sewing continues.
Keep the approved combination intact during this confirmation. Record any new thread lot or needle product. Decide whether that change belongs in the same assessment. Otherwise the successful sample and a later problem may seem to have identical inputs when they do not.
Review the retained samples from the start, middle and end under the same conditions. Where the concern is a particular operation, make sure that operation is represented throughout. A longer length of an easier seam is not a substitute for evidence at the difficult assembly.
Release an instruction that another shift can use
Name the complete thread and needle products in the final record. List each thread position, the fabric lot and the seam assembly. Include the settings actually used, the test methods, results, sample sizes and the person authorised to approve the choice.
State the limits of the approval. A successful combination for one operation does not automatically cover every seam on the style. A new finish or added layer can change the tested conditions. So can a substitute thread or a converted machine.
Keep the approved sample and record linked to the current product revision. ApparelDiary's apparel PLM guide explains how those details can stay connected. Check any later substitute against the recorded requirements. A remembered packet colour is not enough.
A good selection is therefore more than a matching pair of numbers. It links a named thread and compatible needle to the actual material stack. The record shows how this combination meets the seam's requirements. The trial makes that relationship visible and gives the next operator a clear basis for repeating it.
Discussion0
Loading the discussion…