Neither designers nor technicians have much power within knitwear companies, nor influence over their decision making processes, but technicians gain some influence by being difficult to replace. The designers' major source of influence is their special knowledge of future developments in fashion and the requirements of the market, though they often have close ties to customers and suppliers. They are often seen as both junior and temporary, and very easy to replace (as there are more young designers than jobs). This makes managers reluctant to invest money, time or effort in training designers, especially in how to use their companies' CAD systems. (Almost all the designers we talked to expressed an interest in more CAD training than they were able to get, even in the biggest and poshest companies where they have most respect and freedom.) Decisions about which knitting machines to buy are made primarily on price and functionality. However companies are very reluctant to change machine manufacturers unless their current manufacturer cannot provide the functionality and service they need, partly because the expertise of their technicians is to some extent limited to one manufacturer. The designers' only influence on CAD system purchasing decisions comes from knowing what functionality the coming fashions will require. This has the potentially important consequence that there will seldom be much pressure to choose more designer-friendly CAD systems.

We have not investigated the causes of the extremely sharp sex divisions in the knitwear industry, as our brief has been understanding designers, but we should make the following comments. The reasons why knitwear designers are almost exclusively female are rather different from the reasons why the other occupations in the industry have such sharp segregation. Male knitwear designers do not have to face strong social barriers to finding jobs, but men wanting to be knitwear designers appear to be rare. The occupations of knitter and make-up worker (seamstress) grew out of traditional crafts with sharp sex divisions in the industrial revolution; the persistence of such division in trades originating in crafts has been well documented in other industries, and attributed partly to a policy of recruiting to fit into existing social groups (Aitkenhead and Liff, 1991).

Knitting machine technician is an occupation requiring mechanical aptitude, and before the days of electronic machines and computerisation it involved a significant amount of work requiring physical strength (especially moving rolls of metal punch cards, which were heavy). The use of the belief that women are incapable of hard physical work, to exclude women from technical occupations, has been documented recently in other industries by Cockburn (1985); but the hard physical work has been eliminated from the work of knitting machine technicians, which is largely programming and tweaking the controls on knitting machines. Cockburn argues that women are actively excluded from technical fields because men working in them regard technical competence as a masculine attribute fundamental to their self-identities as males. We have little direct evidence on which to base comments on the applicability of this view to the knitwear industry, and none on companies' recruitment policies for technicians. But we conjecture on the basis of our conversations with technicians that a more important factor influencing the sex division is the unconscious bias produced by managers' male images of typical or successful technicians. Webb (1991) remarks that 'It is practically not possible to make an absolute distinction between the job and the job occupant: job requirements are defined in terms of current job holders, including gender.' When asked the question "why aren't there any female technicians?", the technicians we talked to replied "I don't know" or "there aren't any", but did not express hostility to the idea. One experienced technician with whom we discussed this issue commented to us that women used to be regarded by people in the industry as incapable of work requiring mechanical aptitude or strength. He had no objection to the idea of women becoming technicians, and knew one female textile technology student who wanted to, but he thought she would not get a chance, not because of such raw explicit sexism but rather because potential employers would regard her as too much of a gamble, because she did not fit their image of a technician, and would be impossible to dismiss if unsuccessful.

5. The State of CAD System Technology for Knitwear

At the moment commercial knitwear designers make almost no use of computers in the earlier stages of design. The extent to which computers are used for design varies widely between design domains, and this is not the place for a review, but we should note some significant parallels. One is that designers concerned with the design of complex three dimensional shapes subject to various constraints, such as industrial designers, notably automotive designers, make very little use of CAD systems (Tovey, 1989, 1992). Automotive designers are male and in a high prestige industry comprising huge rich companies, so to some extent they constitute a control for the social and economic factors we describe in this paper (though the CAD systems used in car design are so expensive that even Ford think twice before buying them). The exception to this pattern is architecture, where shape design is done using CAD systems; but architecture is largely about the representation and manipulation of simple geometrical shapes, which is relatively easy for a computer to do, and a huge amount of research has been devoted to computerising it (for example Bijl, 1989; Liu, 1991). Perhaps a more significant parallel to industrial design is that both industrial designers and knitwear designers hand over their projects for others to do the detailed technical specification; CAD systems are used extensively by designers in fields where they carry projects from initial conception to detailed technical development, such as architecture, textile design (Devane, 1992) and typographic design (Hewson, 1994).