Nevertheless the knitting machine programming systems are being developed into true CAD systems with features that make them useful for supporting design. All three manufacturers are working on automated programming techniques and increasingly sophisticated ways to model and display the output of the knitting machine, which already include three dimensional representations of what a knitted stitch structure will look like. The development of these display facilities creates the potential for a redivision of the design task between the designer and the technician, a process that has begun at the companies where the designers enter Jacquards. (While knitwear design might benefit from the techniques being developed for computer supported cooperative working (CSCW), this would require a further shift in the conceptualisation of the design process, from one in which the technician takes over from the designer, to one where they work in parallel. We think this is still several years away.)

Despite this rapid technological progress, the developers of knitting machine CAD systems have had very little interest in supporting designers until very recently: system developers at all the three major companies we have talked to told us that their companies regarded supporting designers as of minor importance or not their business. One who thought his company should pay some attention to designers said his was a minority view. One objective ground for this prevailing attitude is the absence of existing demand for designer-friendly CAD systems. In February 1994 Shima Seiki introduced a new version of their CAD system with improved automatic programming facilities and a cutting pattern construction system, which they are marketing as targeted at designers, though it is a linear development of their previous systems and designed primarily for technicians. We found it still very directive in imposing a style of working on the user. Stoll have also started marketing their most advanced systems as having designers as their main user

6. CAD Systems in Industry

Economic factors severely limit designers' access to the existing CAD systems in knitwear companies, and so indirectly limit the demand for designer-friendly systems. The most important factor influencing the use of CAD systems for knitwear, apart from the state of the technology, is simply that the CAD systems are very expensive. For example, the Stoll systems start at 16 000, though additional systems cost 13 000 for the hardware, with free copies of the software, and go up; employing a newly graduated designer for a year is cheaper. The cheapest Shima Seiki CAD system costs 32 000 (which is the cost of a small house in Leicester, where much of the British knitting industry is based), with no site licences or options to buy the hardware separately. The most elaborate state-of-the-art systems, which are the ones with the features most useful to designers, cost up to 134 400. Neither companies nor colleges can afford to provide up-to-date systems for tasks that are not plainly essential; and only the colleges with specialist courses have knitting machines with CAD systems at all. Using CAD systems to design garments, rather than just to program knitting machines, can only be economically viable if the designers' increase in productivity pays for the depreciation costs of the machines they use; this will remain a vital consideration for the forseeable future.

For knitwear companies, investment in giving designers training and access to CAD systems is not clearly essential, even if it might bring benefits, so companies short of money (which most are) will not pay for it. Thus designers often have to content themselves with the technicians' hand-me-downs, which are inevitably less user-friendly and less well-suited to supporting design; they only get access to up-to-date systems when they aren't needed for anything essential. Designers in industry report that they find it very hard to get themselves sent on training courses; this is influenced both by the lesser urgency of giving training to designers compared to technicians, and by the expectation that designers will not remain with the company long enough to return investment in training them; as we have noted earlier, designers work much of the year under enormous time pressure, making it impossible to release them for courses. There may also be sexual discrimination: Hammond (1986) and Colwill and Josephson (1983) report that women find it harder than men in the same occupations to obtain company sponsored training; however we have no easy way to separate this from other factors. However we should add to this pessimistic picture the point that companies' attitudes vary widely, and the knitwear industry as a whole is a moving target; some companies are training their designers on CAD systems and experimenting with computerised design.

As a result of these economic factors, it is very difficult for most designers to make as much use of the CAD systems as they could, which limits the redivision of work the CAD systems are beginning to make possible, and limits the commercial demand for designer-friendly systems. Moreover, the difficulty designers have in gaining training and practice with CAD systems serves to prevent them from gaining the benefits to be had from a greater understanding of the potential and the limitations of their companies' knitting machines.