Sinclair and the 'Sunrise' Technology: The Deconstruction of a Myth (15 page)

The enormous drop in the number of components used in the new machine and a significant fall in price earned Sinclair a justifiably favourable press when the ZX81 was launched. Although the machine’s hidden hardware improvements were to be an important factor in the commercial success of the new machine, it was the improved BASIC that captured the attention of the industry and ultimately the consumer. The reason that the ZX81 was ‘in every way a real computer’ was that Nine Tiles programmers had managed to expand the ZX80 BASIC to incorporate most of the facilities required for practical programming and calculation purposes.

Steven Vickers joined Nine Tiles in January 1980, and learned that his first priority was to improve on the ZX80’s mathematical capabilities. Sinclair realized that if his new machine was to be promoted as an educational tool for the home and the school, it must offer a broad range of trigonometric and floating-point functions. (The ZX80 could deal only with whole numbers.) Sinclair’s brief to his contract programmers was to use the extra 4K of ROM offered by the ZX81’s hardware to develop the machine’s math pack and improve its input/output facilities.

The creation of the ZX80’s software was essentially a one-man operation. It was written by John Grant, who developed the code in consultation with Sinclair Research. As far as the ZX81 was concerned, the work was a division of labour, with Steven Vickers undertaking the expansion of the BASIC, and Grant taking care of the software which handled the computer’s hardware.

One of the most impressive features of the ZX81 is its editing capabilities, which make the alteration of existing program lines extremely straightforward. Given that the majority of owners of the ZX81 were newcomers to computing, the ability to correct errors simply and efficiently was soon recognized as a major selling point. Another of Grant’s bright ideas ensured that only a program line that made sense to the computer could be actually incorporated in a program. Most early micros notified the programmer of such an error only when a program was run, but all the ZX computers indicated any mistakes as soon as an attempt was made to enter the line into the machine. Once again, this thoughtful innovation was a godsend to neophyte programmers.

Most of the ZX81’s software had been completed by the autumn of 1980. The rest of the year was devoted to tidying up the loose ends and writing the manual for the machine. For some reason, Steven Vickers’s documentation, which introduced the ZX81 and its BASIC, received decidedly mediocre reviews in the computer press. This is somewhat surprising, since for the most part the manual offers an adequate overview of the machine and the essentials of BASIC programming, and is far superior to most of the documentation of the day. Certainly Vickers’s text is a considerable improvement on the deficient one supplied with the ZX80 (which, incidentally, was written by Hugo Davenport, later to become Research’s director of engineering). The most consistent gripes about the ZX81 manual appear to centre around Vickers’s slightly eccentric prose, which in retrospect seems a small price to pay for an above-average introduction to computing. Since the provision of accurate and accessible documentation continues to be one of the most conspicuous shortcomings of the microcomputer industry, it’s worth considering Max Phillips’s view of Vickers’s work in his 1983 reassessment of the ZX81:

The ZX81 manual is Steven Vickers’s BASIC tutorial and, given the number of ZX81s and Spectrums sold, must now be one of the classic texts on BASIC. It does a reasonable job and sensibly provides lots for the reader to do. It’s quite honest about the dear thing’s shortcomings and provides hints and tips for ways round them... Best of all, the manual is complete and comprehensive. There’s some fairly advanced and often undisclosed information in there. The beginner won’t understand it for a long time but if he or she learns some more advanced ideas, the manual is ready for them. (What Micro?, April 1983.)

Given that Sinclair had made it clear that he would have the manual rewritten if the ZX81 proved a significant success, it is a credit to Nine Tiles that such improvements proved unnecessary in the light of Vickers’s work. As far as John Grant is concerned, disputes over the ZX81 and Spectrum manuals are one of the main reasons why the software company no longer works for Research. According to Grant, Sinclair had personally suggested that royalties would be paid in respect of both manuals, which, given the vast quantities of machines sold, would have amounted to a tidy supplement to the software fees. In the event, the deal was never formalized, the royalties never materialized and the Spectrum was the last Sinclair project in which Nine Tiles participated. With something approaching grim satisfaction, John and Kate Grant point out that the QL was the first Sinclair computer to be developed entirely in-house, and the first to suffer from serious software failures.

Apart from the completion of the manual, the other major software task left to Nine Tiles was the code that handled output to the new Sinclair printer, which was due to be released at the same time as the computer. Once again finding inspiration in the States (which boasted a number of machines of similar design), Sinclair elected to produce a non-standard thermal printer for the ZX range. With many microcomputing systems, the cost of the printer can outstrip that of the computer itself. Part of the reason for this is that printers normally have to produce printout of a quality suitable for wordprocessing, which makes it impossible for manufacturers to come up with hardware retailing for much less than £250. However, as far as the ZX81 was concerned, the machine’s quirky keyboard and restricted storage capabilities precluded any serious possibility of wordprocessing; the main reason that anyone would want to use a printer with it was simply to obtain readable program listings for reference purposes. The Sinclair ZX Printer used black paper with an aluminium coating and printed thirty-two characters to a line. The mechanics of the printing process were extremely simple. Two styluses were mounted on a belt that moved across the paper. A small electric charge was passed from the styluses, which burnt away the aluminium coating and allowed the black paper underneath to form the appropriate character. The design was crude but reasonably effective, and allowed Sinclair to bring the machine on to the market in November 1981 at the temptingly low price of £49.95. The printer was manufactured at the trusty Timex plant in Dundee, with production kicking off at around 5000 units a month. Although the ZX Printer was far from robust (its output rapidly deteriorating with use), and used relatively expensive paper that never seemed to match the quality of the original roll supplied with the machine, an absence of economic alternatives ensured healthy sales.

In the final analysis, it was the ZX Printer that was behind an infamous error in the ZX81’s software. Although time and the QL have diminished the significance of the ‘square-root bug’, it generated so much correspondence in the computer rags of the day that no history of the machine would be complete without a rendition of the saga. Today, his place in microcomputing history secure, Steve Vickers cheerfully cops the blame for the malfunctioning square-root function. We’ll let a contemporary report in Datalink (21 September 1981) set the tone of the story:

Publicity building up around the fault in the Sinclair ZX81 is attracting the attention of users to the bug ... And to the fury of both Uncle Clive and the 300-odd people who bought the new machines at the PCW show, a whole new bunch of faulty machines has been released ... Sinclair told Datalink that up until the show, only ‘around 300’ people had complained of having faulty systems (they show the square root of 0.25 to be 1.3591409).

John Grant offers a plausible if not necessarily excusable explanation for the bug. As we have already established, Nine Tiles had completed most of the ZX81’s software by the autumn of 1980. However, before the ROM could be finalized, Grant had to write the software to drive the printer, and Jim Westwood and the hardware boys at Research were proving slow in coming up with the goods. When prototypes finally turned up in October, it was discovered that the existing ROM required relatively dramatic restructuring to incorporate the printer code. Grant maintains that it was this process that was the root cause of Vickers’s error.

The square-root saga would be little more than a Trivial Pursuit item for micro addicts were it not for the company’s handling of the affair. Rather than rectify and apologize, Sinclair attempted to minimize and cover up. Once again, we’ll let Datalink (ibid.) chronicle Sinclair’s response to complaints:

‘It really isn’t true that we are deliberately supplying faulty machines in the hope that most users won’t spot [the bug] ... There is a simple test which detects the fault, and Timex, our manufacturing subcontractor, normally test every machine before shipping it.’... He accepted that some customers were having difficulty getting replacements.

In spite of such assurances, it’s difficult to accept that Sinclair Research made much of an effort to halt the shipping of faulty ROMs. The first documented account of the bug appeared in an issue of Datalink dated 17 February 1981. By December of that year, a quarter of a million ZX81s had been sold, and faulty machines were still being reported. It wasn’t simply that the machines were malfunctioning, but that Sinclair Research was far from prompt in replacing returns. Such a situation, coupled with the fact that people were reportedly having to wait up to three months for the delivery of their machines, fuelled the growing suspicion that the company had very little interest in keeping the customer satisfied.

*

At the beginning of this chapter, we noted that there were two external factors capable of significantly influencing the success of the ZX81. The first of these was high-street distribution and, fortunately for Clive, a key figure in one of the country’s major chain stores was preoccupied with much the same product lines as Sinclair and his company. At the end of the 1960s, W. H. Smith & Son was suffering a crisis of identity. No longer everyone’s favourite bookseller, the stores nevertheless retained a stuffy, almost stagnant market image which inhibited their capacity to diversify effectively. A more modern, up-market image was required, and in 1973 the company hired John Rowland as its marketing development manager. Rowland’s brief was to investigate consumer electronics and see what elements of this growing market could be exploited in Smith’s leap into the modern world.

Rowland’s first move was to introduce audio and photographic products into the stores and, encouraged by the success of these new lines, he let Boots suffer the brunt of the fall in calculator prices before taking on the new generation of smaller, cheaper machines. It’s worth stressing that although the move into consumer electronics was both successful and necessary for the company, in the early days W. H. Smith found it difficult to adjust to the unstable prices and rapid obsolescence of the new products. In retrospect, however, this period must seem like the calm before the storm when one considers the chaotic price wars and unpredictable trends of the microcomputer boom that was to come.

Given the prohibitively high price of most personal computers at this time, it’s not surprising that it was computing publications rather than hardware that provided W. H. Smith with its first taste of a promising new market. At the end of the 1970s, there were very few UK-based home-computer publications, and so the demands of a growing army of enthusiasts were satisfied by relatively high-priced US imports. In keeping with Smith’s new image, Rowland felt it appropriate to create ‘computer corners’ in a number of key branches. At such sales points would be gathered magazines like Personal Computer World, Byte, Dr Dobbs and Interface Age along with the dozen or so home-computing books that were all the publishing industry had to offer enthusiasts in those days. The high point of these displays was the handful of Commodore PETs which generated colourful graphics to draw in the punters.

Smith’s computer corners were something close to an overnight sensation, but the company still needed a hardware product if it was successfully to milk this untapped market. Rowland approached Sinclair Research with a view to conducting a marketing experiment with the ZX80. Sinclair recommended that the chainstore wait a few months and enter the world of microcomputers with the new, improved ZX81. Serious interest from a prestigious chain such as W. H. Smith was the break for which Sinclair’s new company had been waiting. Although the ZX80 had been enormously successful, sales of the machine had peaked by the time Rowland made his approach and it was clear that the mail-order market would never be capable of generating sufficient revenue to fund the vehicle and television projects. In addition, by the 1980s Sinclair had become uncomfortable with the ‘
fly-by-night image
’ (Designer, July 1982) of the mail-order manufacturer, and an association with a major retailer offered the opportunity to add a touch of stability and respectability to the corporate image.

Although by the end of 1980 Research had yet to produce a working prototype, Rowland was sufficiently impressed by the ZX81’s specifications to bide his time and await the arrival of the new machine. In January 1981, Sinclair turned up in Rowland’s office with a clay model ‘shaped like a wedge of cheese’, and a deal was signed which enabled W. H. Smith to retail the ZX81 ‘
on an exclusive basis for ... around six months - certainly up until Christmas 1981. After that [the company] would sell it on an exclusive basis for as long as it could sell all the machines [Sinclair Research] could produce.
’ (Interview, 18 October 1985.)

To his credit, Rowland stuck his managerial neck out with the ZX81 and came up trumps. Although confident that the Sinclair machine was precisely the product for which the market was waiting, the dismal response from his major branches could hardly have been encouraging. Presented with the unprepossessing sight of a pre-production ZX81, sceptical buyers around the country suggested launch orders of around 10-15 units per branch. Had Rowland adhered to the unwritten edict of corporate convention, such an overwhelmingly negative reaction would have ensured that the microcomputer experiment was shelved. Fortunately for everyone concerned, he chose to discount his colleagues’ caution and placed the ZX81 at the centre of a major microcomputing promotion.