The Science Behind Your Gauge Swatch: Loop Length, Relaxation, Take-Down Weights, and Why Rows Never Match

Every knitter has been told to wash the swatch, measure in the middle, and not to stretch it. Few have been told why. The reasons are not folklore: they were measured in textile laboratories in the 1950s and 1960s, and the results have held for every generation of yarn since. This is the long version of that research, with the parts that change what you do at the needles or at the machine. If you only want the rules, skip to the end.

1. One number decides the size of your fabric

In 1959 D. L. Munden published a study of plain-knit (stockinette) wool fabrics in the Journal of the Textile Institute. His question was simple: what determines how many stitches and rows a relaxed fabric has per centimetre? The answer surprised the industry. Not the yarn thickness. Not the machine. Not the tension the yarn was fed at. Only one thing: the length of yarn in one loop, written l.

stitches per cm  =  Kw / l
rows per cm      =  Kc / l
stitches per cm² =  Ks / l²

Kw, Kc and Ks are constants. Munden found them to be the same for thin yarn and thick yarn, for coarse machines and fine ones, for loose fabric and dense. For wool after a wet relaxation they are roughly Kw = 4.1, Kc = 5.3, Ks = 21.6, with l in centimetres.

What this means in practice: your needle size or tension dial does exactly one thing. It sets l. Everything you measure afterwards follows from it. Two knitters with different yarns and different machines who produce the same loop length get the same gauge after washing. A thicker yarn at the same loop length gives the same stitch count, only a denser, stiffer fabric.

You can measure l yourself in two ways:

• Unravel. Pull out one row of a known number of stitches, measure the yarn without stretching it, divide by the stitch count.
• Weigh. Weigh the swatch, look up the yarn's metres per 100 g, compute the total yarn in the swatch, divide by the number of stitches in it (stitches per row × rows).

Worked example: a swatch of 40 stitches × 56 rows weighs 9.2 g in a yarn of 250 m per 100 g. Yarn in the swatch: 9.2 / 100 × 250 = 23 m = 2300 cm. Stitches: 40 × 56 = 2240. Loop length: 2300 / 2240 = 1.03 cm. Munden then predicts 4.1 / 1.03 = 4.0 stitches per cm and 5.3 / 1.03 = 5.1 rows per cm after washing, so 40 stitches and 51 rows over 10 cm. If your tape says something far from that, the swatch is either not relaxed or not wool. Either way, the loop length is the one property of your fabric that does not depend on how you washed it, how you dried it, or what mood you were in.

2. Rows and stitches are never "square", and no dial will make them so

Divide Munden's two constants: Kc / Kw ≈ 1.3. That is the loop shape factor: in relaxed stockinette there are about 1.3 times as many rows per centimetre as stitches, because a knitted loop is wider than it is tall. Twenty-two stitches over ten centimetres come with about twenty-nine rows, not twenty-two.

The Soviet knitting-theory school reached the same figure from the other side. A. S. Dalidovich's geometric loop model (Fundamentals of Knitting Theory, 1970) describes the loop as arcs and straight legs and derives a stitch-to-row density ratio of about 0.8 for stockinette, which is simply 1 ÷ 1.3.

The practical consequence is one that pattern instructions rarely admit. Because both stitches and rows are proportional to 1 / l, changing the needle or the tension dial moves them together. If your stitch gauge matches a pattern but your row gauge does not, no needle size will fix it. The ratio is set by the relaxed loop, not by the tool. Match the stitch gauge, measure the rows honestly, and use a pattern that recalculates from your row gauge rather than assuming the designer's. That is exactly what a made-to-measure calculation does, and why it asks for both numbers instead of one.

A note on the "difference" rule

Some machine-knitting schools teach a different target: rows per cm minus stitches per cm should be about 1.2–1.4. Compare it with the ratio and the rule only agrees with the laboratory at fine gauges:

The rule was evidently calibrated on fine cone yarn at 35–40 stitches per 10 cm, where it is correct. Applied to a worsted-weight fabric it demands a loop shape that relaxed stockinette does not have, and the only way to hit it is to measure the swatch stretched lengthwise. A ratio is unit-free; a difference changes when you switch from centimetres to inches. Use the ratio.

The shape factor is also a lie detector. A ratio below 1.15 or above 1.5 in stockinette almost never describes a real fabric. Too few rows usually means a counting error. Too many rows means the swatch was measured stretched lengthwise, typically on the machine bed or straight off it, before the fabric had a chance to recover.

3. "Relaxed" is a defined state, not a feeling

Munden measured his fabrics in two states, and the numbers differ:

• Dry relaxed: the fabric lies flat on a table with no load for a day. Stitches per centimetre about 8 % fewer than after washing, rows about 6 % fewer.
• Wet relaxed: soaked in room-temperature water for a day, squeezed in a towel without wringing, dried flat without stretching. This is the state the constants above describe.

Later work (J. J. F. Knapton and colleagues, 1966 onwards) added a third: fully relaxed, after machine washing and tumble drying repeated until the dimensions stop changing. It is the closest to a garment after six months of wear.

Fabric fresh off a knitting machine is in none of these states. The take-down weights hold it stretched lengthwise, so it shows fewer rows and more stitches per centimetre than it will ever show again. Fabric off hand needles is closer to dry relaxed but still not there. This is the whole reason for the boring advice: a swatch measured before relaxation is a measurement of a state your sweater will never be in.

A home protocol that reproduces the wet-relaxed state closely enough: soak 20–30 minutes, squeeze in a towel, dry flat, measure the next day. Then treat the finished garment the same way.

4. What take-down weights actually do

This is the longest-running argument in machine knitting. One tradition hangs kilograms of weights on a standard-gauge machine; another knits with the cast-on comb alone and adds weight only when the machine drops stitches. Both camps produce sweaters, so who is right? The theory answers the question, and the answer is not "somewhere in between".

The primary job of take-down is mechanical, not dimensional. As a needle rises, the old loop has to slide down the stem and open the latch; as the needle descends, the old loop has to be held below the sinker level so the new loop can be pulled through it. Something must hold the fabric down against the needles. On industrial machines that is done by holding-down sinkers or take-down rollers. On a Japanese domestic flat-bed machine, the gate pegs between needles and the sinker plate on the carriage do most of this job on the main bed, so a comb and a little weight are enough. The old Soviet flat-beds had no such support, so the fabric itself had to be pulled hard enough to clear every latch on every pass. The kilograms come from those machines, not from the fabric. Between the two beds of a ribber there is nothing to hold the fabric, which is why ribbing needs weights on any machine.

The secondary effect is a small change in loop length, through "robbing back". Knapton and Munden showed in 1966 that a loop does not keep the full length the stitch cam pulls: as the needle starts to rise again, tension in the new loop drops and the neighbouring loops, still under tension, pull some yarn back through the needle. The loop ends up a few percent shorter than the cam geometry. How much is robbed back depends on the balance of forces: higher yarn input tension robs more (shorter loops, tighter fabric); higher take-down tension robs less (longer loops, looser fabric). Machines with positive yarn feed remove the effect entirely by metering the yarn; domestic machines have no positive feed, so the effect is real but small, a few percent of loop length.

What follows from this:

• Take-down does not change the shape of the relaxed loop. After washing, the rows-to-stitches ratio is the same whether you knitted with 200 g or 2 kg. The heavy-weight camp is not producing a different kind of fabric; it is producing slightly longer loops at the same dial number.
• The same dial number gives different loop lengths to different knitters, because their take-down, their yarn tension mast setting and their yarn all differ. This is why "knit at tension 6" is not transferable between people and why the swatch, not the dial number, is the unit of exchange.
• Consistency matters far more than magnitude. The classic horizontal band across a machine-knit piece comes from re-hanging the weights: a section knitted at one take-down, then another at a different one, gives two loop lengths in one piece. The swatch must be knitted with the take-down you will use for the garment, and the take-down must stay the same from hem to shoulder. Moving weights up regularly, in small steps, is the practical way to keep it constant.
• Use the minimum that knits without errors. Edge weights stretch the edge loops more than the centre, producing flared or wavy selvedges; heavy weights on a fine yarn raise the yarn tension at the knitting point and cause breaks and dropped stitches on the ribber. The theory gives no reward for extra weight beyond error-free knitting.

So the debate resolves like this: the kilograms were a workaround for machines without gate pegs and sinker plates; on a machine that has them, one comb and a stable, light weight is not laziness, it is the setting the theory prefers, provided it is the same on the swatch and on the garment.

5. Modern yarns: what still holds and what does not

Munden worked with wool. The constants describe the geometry of a relaxed loop, not the chemistry of a fibre, so the loop shape factor of about 1.3 turns out to be remarkably stable across fibres. What varies between fibres is whether the fabric reaches equilibrium at all:

• Wool recovers on its own when wet. The classic case.
• Cotton does not reach equilibrium from soaking alone; Knapton showed in 1975 that its constants drift with fabric tightness and that it needs tumble drying to settle. Expect growth.
• Acrylic and blends are fixed by heat. A steamed acrylic fabric stays in the state it was steamed in, and the constants apply only until the next steaming.
• Superwash wool does not felt, but it relaxes further than untreated wool and grows more in length under its own weight.
• Viscose, bamboo, silk grow lengthwise after washing and under hanging load.

None of this can be predicted from the label. It can be measured from a swatch that was washed and dried like the garment and, for heavy or slippery fabrics, hung with a small weight for a day to see how much it grows. The swatch is the model; the label is a hint.

6. Ribbing and other structures

Everything above is about stockinette. Knapton and Munden measured 1×1 rib, 2×2 rib and half-cardigan in 1966 and found that the same law holds: dimensions follow the loop length alone, but each structure has its own constants and its own loop shape. A rib loop after relaxation is much narrower than a stockinette loop because the wales pull together, so a rib swatch gives many more stitches per centimetre at the same loop length, and the rows-to-stitches ratio of stockinette must not be applied to it. Two rules survive: the loop length is still the invariant, and a ribbing gauge has to be measured on a ribbing swatch, relaxed, never inferred from the main fabric.

7. Thickness, tightness, and what "doubling the yarn" really does

Munden also proposed a tightness factor, K = √tex ÷ l (with l in millimetres and tex the yarn's weight in grams per 1000 metres). Values around 1.4–1.5 give a fabric that drapes and wears well; below 1.0 the fabric is loose, skews and snags; above 2.0 it is a board. Dalidovich's loop modulus (loop length divided by yarn diameter, typically 17–25 for wool stockinette) is the same quantity from the other side.

Yarn diameter is not printed on any label, but it follows from the tex:

d (mm) = 0.0357 × √(tex / δ)
δ = bulk density of the yarn, g/cm³: about 0.8 for wool, 0.9 for cotton, 0.75 for acrylic

Two facts about thickness that surprise most knitters:

• Thickness follows the square root of weight. Two strands of the same yarn held together are 1.41 times as thick as one, not twice; three strands 1.73; four exactly twice.
• On a cone, "2/28" means two plies of metric count 28 (28 metres per gram each), 71 tex in total. A 50 g skein of 125 m is 400 tex. Convert everything to tex, add the strands, and the diameter formula works for any mixture. For a mix of fibres, weight the density by mass.

8. How much yarn: consumption from loop length

Dalidovich's textbook contains the other half of the loop-length law, the one hand-knitting books never mention. If every stitch contains l of yarn and you know how many stitches a square metre holds, the weight of a square metre of fabric is arithmetic:

grams per m²  =  l (m) × stitches per m² × tex / 1000
stitches per m²  =  (stitches per 10 cm × 10) × (rows per 10 cm × 10)

Example: 20 stitches and 28 rows per 10 cm, loop length 20.5 mm (which is what Munden gives for that stitch gauge in wool), 400 tex yarn. Stitches per m²: 200 × 280 = 56 000. Yarn per m²: 0.0205 × 56 000 = 1148 m. Weight: 1148 × 400 / 1000 = 459 g per m². A medium pullover is about one square metre of fabric, which is why it takes about 450–500 g of worsted-weight wool. A pattern that knows the area of its pieces can turn this into a yarn estimate for your size and your gauge, instead of "buy one extra ball".

9. What to do with all this

1. Wash and dry the swatch like the garment. Do not measure on the machine or off the needles.
2. Knit the swatch with the same yarn, the same number of strands, the same tension mast setting and the same take-down as the garment. Then keep the take-down constant from hem to shoulder.
3. Match the stitch gauge. Report the row gauge honestly; let the pattern recalculate the rows.
4. If rows ÷ stitches in stockinette is outside 1.15–1.5, distrust the measurement before distrusting the yarn.
5. When the stitch gauge is off by a known percentage, the loop length must change by the same percentage. Re-swatch rather than extrapolate, because a dial step is not the same size on every machine or for every knitter.
6. For cotton, viscose, superwash or heavy fabrics, add a hanging test: the swatch will tell you what the sweater's own weight is going to do.
7. Weigh the swatch. With the yarn's metres per 100 g it gives you the loop length, and the loop length gives you both a check on your measurement and an estimate of the yarn you need.

Sources

• Munden, D. L. (1959). The geometry and dimensional properties of plain-knit fabrics. Journal of the Textile Institute, 50, T448–T471.
• Knapton, J. J. F., Munden, D. L. (1966). A theoretical study of loop formation and yarn tensions in weft knitting (parts I and II). Textile Research Journal, 36.
• Knapton, J. J. F., Ahrens, F. J., Ingenthron, W. W., Fong, W. (1968). The dimensional properties of knitted wool fabrics. Textile Research Journal, 38.
• Knapton, J. J. F. et al. (1975). Dimensional properties of cotton plain-knit fabrics. Textile Research Journal, 45.
• Далидович, А. С. (1970). Основы теории вязания. Москва: Лёгкая индустрия.
• Кудрявин, Л. А., Шалов, И. И. (1991). Основы технологии трикотажного производства. Москва: Легпромбытиздат.
• Spencer, D. J. (2001). Knitting Technology, 3rd ed. Woodhead Publishing.