Connective systems for machine knitting: procedures, data, software, and curriculum for the Shima Seiki lab

Nicole M. Germack

doi:10.17918/00011541

To design knitwear on an industrial knitting machine, one person has to operate several expert systems at once: the machine and its programming software, yarn count mathematics in five unit systems, the behavior of knit structures, color, costing, and finishing. Each system is complete on its own terms, and none of them connect. The knowledge needed to move between them is scattered across manuals, cone labels, reference tables, machine settings, and people, so every act of making requires a chain of translations that the maker performs by hand and that no record preserves. In a university lab, where a student gets one or two terms with the machine, this fragmentation is the binding constraint, setting what can be learned, made, and researched. This thesis is the record of a systemic response to that problem, built over three years in Drexel University's Shima Seiki knitting lab. The response has five connected parts, built in order. I learned the machines and wrote their operating knowledge into standard operating procedures: yarn counts and their conversions, yarn selection by gauge, stitch settings, the pattern workflow from drawing to machine file, and consumption and costing. I ran those procedures across a study of 117 documented knit samples, 15 yarns crossed with 6 knit structures on the 12 gauge machine, each sample washed and measured, 84 of them weighed. The procedures and the records then went into a PostgreSQL database whose tables mirror the decisions a knitter makes and whose rules answer by query what experts answer from memory. On that database I built NIKY, a web application that replaces the paper tech sheet with a searchable record and gives students working calculators for yarn counts, gauge planning, costing, and combining ends of yarn. I then taught the lab's two machine knitting courses through this system, so the technical record students already produce for a grade is the record the database stores. The 117 study records are live in the lab's production database. The study returned numbers the lab never had. Every knitter knows that structure changes a fabric; no one at this machine could say by how much, and the study's measurements now answer that question directly. Knit on the same needles, a rib finishes at less than half the width of a double jersey, and the heaviest structure weighs about twice as much per square meter as the lightest. Each structure's contraction from needle bed to washed fabric is now a measured ratio, so a target size converts to a machine program by arithmetic instead of repeated trial runs. One working assumption failed outright: the yarn's stated thickness, the number knitters use as shorthand for how fine a fabric will finish, predicted nothing in this set. The infrastructure has also proved generative. The archive grows as a side effect of coursework, because the record a student submits is the record the database keeps. New work has already spun out of it: an image-to-stitch conversion tool, a dye experiment whose samples entered the archive on arrival, and a second relational archive, for pigments in historical paintings, built on the same method. The result is a working model for a lab whose knowledge accumulates: procedures anyone can read, data anyone can query, tools built for novices, and a record each contributor controls.

Connective systems for machine knitting: procedures, data, software, and curriculum for the Shima Seiki lab

Files and links (1)

Abstract

Metrics

Details

Connective systems for machine knitting: procedures, data, software, and curriculum for the Shima Seiki lab

Files and links (1)

Abstract

Metrics

Details

Drexel University Social media