Much Ado About Knotting

Abstract: This article will provide a new approach to estimating local and global knot density of a knotted pile textile. It will introduce random sampling as a way to provide rigor, precision, and accuracy. Applying this approach to two weavings from The Textile Museum Collection clearly demonstrates that multiple samples of a single nomadic carpet’s local knot density do not all have the same value—rather, they vary quite significantly (often by 30 to 45 percent of their average value) over the textile. Detailed analysis of the sample data suggests that local knot density is an example of what mathematicians define as a random variable, with the probability for it having any particular value determined by a special probability distribution function—the well-known normal distribution, the famous “bell-shaped curve.” For a technical analysis to fully describe a regular knotted pile weaving’s estimated global knot density, it is necessary to know both the mean knot density and the confidence interval or, equivalently, its “margin of error” or “error bars.”