For uniformity purposes, wash testing is typically done using reference "monitor" textiles from specialist mills. These fabrics are made with consistent materials and processes. This creates testing series of excellent statistical reliability - if the tests were repeated with different batches of the textiles, the results could be expected to be nearly identical.
The problem with this approach is that it's not statistically valid - that is, the results with monitor textiles don't have a consistent relationship to results you'd see with the textiles consumers actually wear and live with.
For this program, I've created eleven 12-pound washloads of everyday clothing and home textiles representative of what people actually wear and live with. I've selected both black and navy blue options to create a washload that looks like how normal consumers do laundry, if they sort with a moderate degree of intensity. To maximize the broad applicability of the test, I've maximized the number of distinct garments or home goods by splitting most garments into smaller pieces. For example, t-shirts are split into a front and back half, pants are divided into sections of legs.
Because full synthetic fabrics like polyester, nylon and acrylic don't lose color in typical wash processes, the selections have been made to bias for natural fiber content, and because the testing in the Phase III OBA/Percarbonate trial will be performed with typical consumer laundry detergents with proteases, animal fiber content has been excluded.
The Color Evaluation page describes in detail how non-cellulosic content in the garment body affects the weighting of scores. Items that are heavily- to fully synthetic are included for comparison, but do not contribute to color retention scoring.
I'm currently preparing the monitors; these include items from such retailers as Walmart, Target, JCPenney, Kohl's, Macy's, Gap, Old Navy and Lands' End, as well as popular graphic tee blanks, scrubs and workwear brands. There will be a complete gallery here shortly.