Every patterned fabric produced before about 1800 was woven on some version of a drawloom, a type of loom equipped with a mechanical memory that stored the pattern and repeated it with each pass of the shuttle. Compared with the simple looms that produce plain cloth, a drawloom is an astonishingly complex machine, and the most elaborate versions - the Chinese and French looms used for silk brocades in the seventeenth and eighteenth centuries - could weave patterns with several thousand independently controlled threads, producing designs of a complexity and fineness that no hand-embroiderer could match. The intellectual lineage of these machines runs directly into the computer: the punched-card mechanism that Joseph-Marie Jacquard added to the drawloom in 1804 became the direct inspiration for the data storage systems of nineteenth-century mechanical calculation.

The basic problem a drawloom solves is how to select, on each pass of the weft, which of the hundreds or thousands of warp threads should be raised above the shuttle's path and which should be left below. Each warp thread is attached, via a heddle, to a cord that can lift it when pulled. On a simple loom, the cords are grouped in perhaps four or eight sets, and a small number of foot pedals controls which set is raised at each pass. This produces simple geometric patterns. For a complex figured fabric, each warp thread must be controllable independently, and some other mechanism is needed to make the selection.

The traditional solution, used in China from at least the Han dynasty and in Europe from the medieval period, was to have a second person work with the weaver. This second worker, called a drawboy or drawgirl, sat above the loom and physically pulled groups of cords as the pattern required. The groupings for each pass were set out in advance on a system of cord loops, and the drawboy worked through them in sequence. A complex silk brocade required several years of preparatory work to set up the cord system, and the resulting pattern could then be produced repeatedly as long as the setup remained intact. The limitation was that a drawloom required two workers, one of whom - the drawboy - was doing skilled but repetitive work for long hours.

The French silk industry of the seventeenth and eighteenth centuries concentrated in Lyon, which became the centre of high-pattern weaving in Europe. Lyon's economy depended on the drawloom, and successive generations of mechanical improvements sought to make the cord-pulling more efficient. The French silk historian Marguerite Deschamps has described a succession of inventions by Lyon weavers through the eighteenth century, each of which reduced the skilled labour required. None, however, solved the fundamental problem that a drawloom needed two workers, until the arrival of the Jacquard mechanism.

Jacquard's contribution was to replace the drawboy with a mechanical selector controlled by perforated cards. Each card represented one pass of the weft, with holes punched in positions corresponding to the warp threads that should be raised. The cards were tied together in a continuous loop and fed through the mechanism in sequence. As each card passed over a grid of pins, the pins aligned with holes could pass through and raise their corresponding cords; the others stayed down. The mechanism could be built to handle several thousand independently controlled warp threads, and it replaced the drawboy entirely. A single worker could now produce the most complex figured fabrics, provided the cards had been properly prepared. The time-consuming setup was shifted from the cord system of the drawloom to the punching of a card set, which had the advantage that a well-made card set could be stored and used repeatedly without needing to be set up again.

The economic effects were considerable. The Jacquard loom reduced the cost of figured silks substantially, though it did not displace the drawloom everywhere - some traditional Chinese workshops continued to use the older technology into the twentieth century - and it allowed a significant expansion of the range of people who could afford patterned fabrics. The human cost, however, was real. The drawboys whose labour had been replaced lost their work, and the skilled weavers who had long been the elite of the industry found themselves in competition with the new mechanisms. The 1830s saw significant unrest in Lyon, with weavers protesting against changes they saw as threatening their livelihoods. Similar patterns followed in other industries as mechanisation spread, and the Lyon weavers' protests have since been read as an early example of the social disruptions that industrial machinery would produce throughout the nineteenth century.

The connection to computing is more than a historical curiosity. Charles Babbage, designing his Analytical Engine in the 1830s, borrowed the Jacquard card directly as the input medium for his proposed machine. The cards would specify the calculation to be performed, as they had specified the pattern to be woven. Ada Lovelace, in her published notes on the machine, explicitly compared its operation to the weaving of algebraic patterns - a comparison that captured something essential about what computers would eventually do. The information-storage principle of the Jacquard card persisted in the punched cards used in twentieth-century tabulating machines and, with modifications, in the paper tape and punched cards of early electronic computers. It was only fully displaced by magnetic storage in the 1960s.

The continued interest in drawlooms is today largely historical. A small number of museums and research centres maintain working examples, and a few textile artists have learned to operate them, producing contemporary figured fabrics with techniques of considerable antiquity. Deschamps has argued that the drawloom deserves recognition not only as a textile tool but as the world's first programmable machine, predating the mechanical calculator by several centuries and setting the conceptual pattern that the computing revolution would eventually follow. The patterned silks produced on these machines are themselves memory devices, recording in thread the decisions stored in the cords and cards that made them. Looking at a seventeenth-century brocade, on her account, is a way of looking at one of the earliest stored programs of human history.