The arithmetic everyone gets backwards
A fabric loses eight percent of its length in the wash. You need ten yards after washing. How much do you buy?
The instinct is ten plus eight percent, which is 10.8. That is wrong, and it is wrong short. Eight percent comes off the length you buy, not off the length you want. Buy 10.8 and you get 10.8 times 0.92, which is 9.94 — two inches shy, which is exactly enough to matter on the last piece.
The correct operation is division. You want a length that still measures ten after losing eight percent of itself, so it is ten divided by 0.92, which is 10.87. The two answers are close at small percentages and diverge fast:
| Shrinkage | Add the percentage | Divide (correct) | Short by |
|---|---|---|---|
| 3% | 1.030 | 1.031 | 0.1% |
| 8% | 1.080 | 1.087 | 0.7% |
| 15% | 1.150 | 1.176 | 2.3% |
| 30% | 1.300 | 1.429 | 9.0% |
At three percent nobody notices. At thirty, which is not unheard of in an unbleached loose weave or a wool flannel, adding the percentage leaves you nine percent of the purchase short.
Measuring the swatch so the number means something
The measurement error is fixed; the square is not. Mark a ten inch square and mis-measure by a sixteenth and you have a 0.6 percent error in a reading that might be three percent. Mark a twenty inch square and the same sixteenth is 0.3 percent. Bigger is better, and the practical limit is the width of the goods.
Mark with something that survives washing, on the fabric itself rather than on a tag, and mark the square with the sides parallel to the grain — one pair along the selvage, the other across it. Measure between the same points before and after, from the outside of the mark to the outside of the mark, and record both directions separately. They will not be the same.
Wash the swatch the way the finished item will actually be washed, including the dryer. A cool gentle wash and a hot wash with a hot tumble are different experiments and give different answers, and there is no way to convert between them. If the item will be washed hot for the rest of its life, test hot.
Length and width shrink differently, and that is the interesting part
Woven cloth is held under tension along its length while it is finished, and it relaxes when that tension comes off in the wash. Across the width there is much less tension to release. The result is that lengthwise shrinkage usually exceeds crosswise, often by two or three times, and a square you marked comes back as a rectangle.
Two consequences follow. The first is that any piece cut on the crosswise grain will not shrink the same way as a piece cut lengthwise, so a garment that mixes grain directions builds a distortion into itself unless the fabric is preshrunk. The second is the usable width: on a 44 inch bolt losing 2.5 percent across, the width comes back at 42.9. If your pieces were 21 inches wide and two just fit across 43 inches of usable width after selvage trim, they do not fit any more, and the yardage doubles.
That step is the reason to test before buying rather than after. The fabric yardage calculator takes a shrinkage percentage as an input and will show you the two-across to one-across cliff directly; the percentages this page produces are the ones to type in.
Diagonals, skew and what washing reveals
Measure both diagonals of the marked square after washing. If they differ, the square has racked into a rhombus, which means the warp and weft are no longer at right angles. Fabric can leave the mill this way, held straight by finish, and the wash lets it relax into the shape it actually is.
Skew cannot be corrected by cutting carefully. If the print is off grain it stays off grain. If the weave is off grain you can sometimes pull the cloth back on the bias, wet, before cutting, and it will hold well enough for a garment. What you cannot do is cut a straight piece from a skewed cloth and expect it to behave, because the grain that provides the stability is not running where the piece thinks it is.
A degree or less is not worth worrying about. Three degrees on a shirt front is visible after two washes as a twisted side seam.
To prewash or not
If the finished item will ever be washed, prewash. Shrinking after cutting distorts pieces relative to each other; shrinking after sewing shrinks the fabric against thread that does not shrink, and every seam puckers.
The cases against prewashing are real but narrow. Fixed upholstery and wall panels never get washed, and washing removes the finish and changes how the cloth handles. Quilting cotton is argued about endlessly: prewashing costs you the crispness that makes accurate piecing easier, and some quilters prefer the slight crinkle that comes from letting a finished quilt shrink once, which requires that all the fabrics in it shrink about the same amount. That last condition is the one to test for, and testing several swatches together is the way to find out.
For the pieces themselves, the seam allowance and cut size calculator turns a finished dimension into a cut size, and the shrinkage factors on this page multiply straight into it. For a quilt specifically, the quilt backing and binding calculator has its own shrinkage field for the backing, which shrinks against a top that may have been treated differently.
Questions people ask
How big should the test square be?
At least ten inches on a side, and larger if the fabric allows. The reason is arithmetic: your measuring error is roughly constant, perhaps a sixteenth of an inch, so it is a smaller fraction of a bigger reading. A sixteenth on a ten inch square is 0.6 percent of the measurement, which is a fifth of a typical three percent shrinkage. On a twenty inch square it is 0.3 percent. Mark the square well inside the swatch so the cut edges, which fray and distort, are not part of what you measure.
Should I wash the swatch once or twice?
Once tells you most of it. The first hot wash releases most of the tension the cloth was finished under, and that is usually the bulk of the movement. A second wash typically adds a fraction of a percent, occasionally more on a loose weave or an unbleached cotton. If the first wash gives a large number, over about twelve percent, wash the same swatch again and measure again before you commit to a purchase, because a fabric that moves that much often has more to give.
Why is lengthwise shrinkage bigger than crosswise?
Because of how cloth is finished. It travels through the finishing machinery under tension along its length, which stretches it, and the wash lets that stretch come back. There is far less tension across the width. Two to three times more lengthwise than crosswise is a common pattern in woven cotton. Knits often behave differently and can even gain width while losing length, which is why both directions are separate measurements here rather than one number.
Can I just add ten percent and be safe?
For a small piece from a fabric that shrinks three percent, yes, and nobody will ever know. The habit becomes expensive in two directions. On a low shrinkage fabric a flat ten percent is fabric you paid for and will not use. On a high shrinkage one it is not enough, because ten percent added is only 9.1 percent of the purchase, and a cloth losing fifteen wants seventeen and a half added. The swatch takes one wash cycle and replaces the guess with a number.
What does it mean if the diagonals of my square are different lengths?
The cloth is off grain, and the wash revealed it. The warp and weft are no longer perpendicular, so the marked square has racked into a rhombus. A small difference, under about half a degree, is not worth acting on. A large one means any piece you cut will have its grain running at an angle to its edges, which shows up later as twisting seams and hems that will not hang level. Wetting the fabric and pulling on the bias can bring the grain back on a plain weave; on a print, the print stays where it was printed.