Draw-In and the Tyranny of Small Numbers

Marcy Petrini

August, 2023


I was wondering whether the various extruded natural fibers (Tencel™, bamboo, etc.) have the same shrinkage. I was looking through the spreadsheets of my weavings, one per project.

Looking for this kind of information is the reason to keep good notes. For every piece I weave, I know the width on the loom, sometimes adjusted to obtain a complete repeat of the pattern; I measure the fabric off the loom; I wet finish it as appropriate for the fiber, and measure after. The last two measurements give me the percent shrinkage.

The first two measurements give me my take-up, as I learn it, or draw-in. Whatever you like to call it, the warp is at a certain width on the loom; with the weft going over and under it, there is a certain amount of loss in the width. This happens also in the length. Because the calculations for the take-up and shrinkage are the result of different processes, I like to keep them separate.

What caught my attention in looking through my files was the variation in the width take-up (or draw-in): 7.4% for the blue and yellow scarf below (which I wrote about in my January blog). Its initial width on loom was 8.1”, adjusted for the pattern repeat.




In contrast, the Swedish lace shawl below had a 35.3” width on the loom, again adjusted for the pattern; its take-up was 4.5%. Granted, both percentages are rather small and acceptable, but somehow, I expected the opposite, as I think it may be a bit harder to allow enough weft in a wider piece to avoid the draw-in.




Was I wrong in my expectations? No, actually, the difference in the percentage is what my husband calls “the tyranny of small numbers.” Making conclusions based on small samples is unreliable, and, in this particular case, the smaller width makes the draw-in look bigger.

Here are the actual widths and calculations:


Width adjusted

Off the loom

Take Up


Yellow and blue scarf





Swedish Lace shawl






My hunch was correct – larger piece, more draw-in, although the comparison is not exactly fair since the structures are different.

I am not drawing any conclusions based on two pieces of weaving! My point is that comparing percentages – and ratios – with relatively small numbers can be deceiving. The smaller difference looks larger because it’s a percentage of a smaller number.

We see this deception in everyday life.

At the ballpark, I have watched batting averages (which are actually ratios) go up and down in what seems an unpredictable way. A 300 batter (0.300 actually) with 100 at-bats, gets a hit, and his average goes up to 307. But if he hadn’t gotten a hit, his batting average would have dropped to 297. Over the course of a season, these fluctuations become much smaller and may be unnoticeable, but I have heard broadcasters talk about these deceiving numbers as if they were reliable.

What about my original reason for digging into my files? I couldn’t arrive at any conclusions because I didn’t have enough pieces woven with one fiber only, since I generally mix fibers.

However, the lesson stands. Looking at the percentages of shrinkage was deceiving, too.

Happy weaving!