Sinclair and the 'Sunrise' Technology: The Deconstruction of a Myth (18 page)

Thus, although the Spectrum boasted an impressive expansion of ZX BASIC, the new facilities were impaired by their inefficient implementation. In short, the execution of a Spectrum program was depressingly slow, as the following reviewer’s comments emphasize:

The BASIC is slow, well, ‘snail-like’ would be a better description, and the standard Benchmark results are given in Table 1. The last test was done with a loop of 100 instead of 1000 as I thought that you might like to read the review before the Christmas holidays. (Computing Today, August 1982.)

Initially, the absence of competition and an inexperienced market meant that the Spectrum’s deficiencies had little effect on its success. Had the machine been developed as an interim product, then its shortcomings would have been defensible. That Jan Jones was hired at the beginning of 1983 specifically to create a Spectrum Super BASIC suggests that Sinclair may have taken Grant’s concerns to heart. That the Superspectrum was abandoned is a symptom of the complacency that lost Research its market lead as the competition hit back with superior products.

The stalwart Jim Westwood was conspicuously absent during the development of the Spectrum’s hardware. The £5m flat-screen investment programme initiated at the beginning of 1981 had apparently brought the company no closer to getting a product on the market. In desperation, Westwood had been taken off the ZX range after the completion of the ZX81 and dispatched to rescue the television. Into Westwood’s shoes stepped the admirably capable Richard Altwasser. That the Spectrum reached the market more or less on schedule is largely a result of the friendship that developed between Vickers and Altwasser, with the latter serving as a buffer between Nine Tiles and Sinclair Research. Having skimped on the machine’s software development, a decision seems to have been taken to be a little more generous with the hardware. A number of reviewers noted with satisfaction that, unlike earlier machines, the micro’s clock circuitry and display were crystal controlled - a fact that contributed to the reliability of the hardware. Others pointed to the tidy internal layout, for which credit must go to the cooperative labours of Dickinson and Altwasser.

The work on the Spectrum’s software took the best part of a year to complete. Although the straight enhancements of ZX81 BASIC were relatively unproblematic, the development of the code that was to handle the various planned peripherals was impeded by the lack of working hardware. After six months shunting between Research and Ferranti, Altwasser finally managed to cobble together a prototype of the Spectrum itself, and by Christmas Vickers had completed the bulk of the software. From this point on the situation deteriorated.

The problems started in February 1982 with financial disagreements between Nine Tiles and Sinclair. The Grants insist that for years Sinclair had been suggesting the possibility of royalties on their work, and when it became clear that these were not going to materialize, they decided to put up their fees. For his part, Sinclair made it clear that the company’s rates were over the top. In a product development impeded by bad feeling, the announcement by Vickers and Altwasser that they were departing to form their own company couldn’t have come at a worse moment. (They went on to trade as Cantab, which was to produce the ill-fated Jupiter Ace, a computer featuring the Forth language.)

With Altwasser gone, for a while hardware development drifted along at Research with no one at the helm. In February, with the Spectrum’s April launch looming and still no sign of completed peripherals, it was decided to produce the incomplete ROM for a limited release. Grant explains the theory behind the strategy:

The original idea was that Research were going to bring out the Spectrum with an unfinished ROM. They were going to make just a very few. They knew that before long ... they’d have the real ROM and anyone who bought add-ons for an early machine would be able to have an exchange ROM. Then it got to somewhere around May or June and they’d sold 75,000 machines, all with the old ROM. They came to the conclusion that the original idea just wasn’t going to be viable. (Interview with John Grant, 8 September 1985.)

Grant resolved this potentially disastrous situation by coming up with the idea of a ‘shadow’ ROM that sat on the add-on card and took over from the Spectrum’s ROM when the peripheral was called into use. This led to the absurd situation in which the Spectrum’s software was still being developed more than three months after the machine’s launch! In effect, the resident Spectrum ROM was to remain incomplete. Whereas the ROMs of the earlier ZX micros were crammed to the hilt, that of the Spectrum boasts 1300 free bytes, which had been reserved for the peripheral software.

Since the Spectrum was the last Sinclair product on which the company was to work, this seems an appropriate point to record the views of Nine Tiles on the Spectrum development. These are summed up in this extract from a letter to Sinclair:

During the last year, the project has been subject to abrupt changes in direction and considerable effort has been wasted. For instance, the way [the network] has been used has been changed several times in March and April of this year. Software was completed in April and we have not yet been told that there is any hardware on which to test it. We feel that there is a need for a more structured approach to the planning of the project with the hard- and software design’s timescales being agreed beforehand by other members of the team. (Letter from John Grant to Clive Sinclair, 12 June 1982.)

Although it is easy to dismiss Grant’s comments as the fruits of resentment, it should be remembered that Sinclair has levelled similar criticisms at the rest of the computer industry (see page 155). Given Sinclair’s track record and the circumstances under which the Spectrum was developed, it is difficult to believe that Grant’s complaints are entirely without foundation.

There’s an ironic postscript to the Sinclair-Nine Tiles saga. After the split with Research visitors to Nine Tiles were puzzled by the sight of staff wistfully toying with calculations involving multiples of 2.5 million. The truth can now be told. It seems that at the end of 1984 Nigel Searle, then managing director of Sinclair Research, found himself grappling with a legal problem in the Far East. Pirate Spectrums were flooding the market and the company was constructing a case that would enable it to sue for breach of copyright. The only trouble was that no one seemed able to lay hands on the document that established Sinclair’s ownership of the Spectrum’s software. With a growing sense of dread, it finally dawned on Searle and Sinclair that as far as anyone could tell the company didn’t actually hold the software copyright! In the panic and bad feeling that marked the closing stages of the micro’s development, no one had got around to asking Nine Tiles to sign over the appropriate pieces of paper. In theory, John Grant’s company may well own the software copyright of the world’s bestselling microcomputer, unit sales of which have now passed the 2.5 million mark. It seems that the company declined Sinclair’s modest offer for the relevant documentation. After all, almost anything multiplied by 2.5 million comes to more than £5000, doesn’t it? Anyway, the 1985 cash crisis interfered with the resolution of this potentially vexed issue, which was presumably sorted out in the course of the sale to Amstrad of Sinclair’s intellectual property rights where they relate to computers in 1986.

Given the untidy conclusion to the development programme, it was hardly surprising that supply problems hit the Spectrum on an unprecedented scale. Following the standard pattern for a Sinclair launch, purchase was at first restricted to mail-order sales, but the company departed from its usual advertising strategy by initially confining its campaign to the pages of the computing publications. Nevertheless, demand was enormous and although the micro was officially launched in April 1982, it wasn’t until June that the first machines began to trickle into the hands of the customers. Never prone to a sense of deja vu, Sinclair was once again reported to have been ‘utterly astonished’ by demand for his new product

By July, Sinclair Research was sitting on a backlog of 30,000 orders. Once production had got into full swing at Timex, Dundee, the manufacturer was pumping out 5000 units a week. Then, in mid-July, just as supply problems were in sight of being resolved, Timex shut down their entire plant for its three-week annual holiday. The backlog hit the 40,000 mark, and customers were told they could expect a wait of anything up to twelve weeks. With memories of ZX81 delays still fresh in their minds, the thousands who paid their money and then waited three months for delivery must have found it difficult to escape the suspicion that they were forward-financing Sinclair production. There were dark rumblings of dissent in the computer journals and news of the crisis spread to the national papers.

Applying a Band-aid to the savaged jugular of his PR, in September Sinclair published an open letter in the computing press apologizing for the delays. He offered money back on demand to those who were fed up with waiting, and a £10 voucher towards printer or printer paper to those blessed with unnatural patience. However, as far as the Advertising Standards Authority (ASA) was concerned, it was a case of too little, too late. For all the good it did the customers, in October the organization upheld complaints against Spectrum ads (which promised ‘28-day delivery’) and berated the company for the ‘appalling delays’ in fulfilling orders.

Given the backlog of orders for the Spectrum, it was clearly important to sustain ZX81 sales, particularly since it was the only Sinclair microcomputer making a showing in the retail outlets. Sinclair was convinced that, ‘The ZX81 will continue to be ideal for the person who wants the lowest possible entry cost into computing.’ To prove his point, at the Spectrum’s launch he knocked down the price of the ZX81’s RAM pack from £49.95 to £29.95, although the price of the ZX Printer was raised to £59.95. Far from cutting back on manufacture of the old machine, the company boldly announced that its ZX81 production target for the end of the year was 150,000 units per month. Despite the delays many first-time buyers chose to hold out for the Spectrum and there was a significant slump in sales of the ZX81. To counter this trend, Research signed up Prism Microproducts to wholesale the machine, extended high-street sales to include the Boots chain and cut the price of the ZX81 to £49.95.

In the frenzy and frustration of the months following the Spectrum’s launch, Sinclair received solace and/or encouragement from an unexpected source. Caught by the wave of micro mania that was sweeping the country, the progressive forces of the Conservative government decided to step up their contribution to the hi-tech revolution. Fired by the success of the Micros for Secondary Schools campaign, the Department of Industry (DI) committed a further £9m with the intention of putting at least one computer in every primary school. Just as pre-production machines were being readied for the reviewers, the Spectrum and ZX81 were tested by the DI to see if they qualified for a place in the programme.

In July 1982, Margaret Thatcher announced details of the scheme to the nation and revealed the products that had been deemed educationally sound - the Sinclair Spectrum 48K, the BBC Model B with disk interface and the Research Machines Link 480Z. With more than 27,000 institutions expected to take part in the scheme, Sinclair Research felt that there was cause for celebration. On hearing the news, a spokesman made it clear that the company believed it could knock the competition out of the running: ‘
We’re very pleased about it. We hope that being the cheapest in the scheme, we will be able to outsell the other manufacturers.
’ Ironically, although Sinclair has been extremely successful in getting his machines into institutions all over the world, the company managed to secure only a paltry 2 per cent of the UK education market. Among the reasons given by British teachers are that the Spectrum’s size makes it too easy to steal and that the machine is simply too fragile to withstand the battering it could expect from hordes of pre-teens.

As the Spectrum looked set to secure Sinclair’s dominance of the UK home-computer market, plans were already well advanced for the company’s assault on the US. In July 1982, Timex launched the TS1000 in the States, so this seems an appropriate point at which to chronicle the relationship that developed between the two companies and outline their American ventures.

We saw in the last chapter how in 1981 Fred Olsen took the first steps in the rescue of his floundering Timex Corporation by diversifying into assembly and manufacture in the UK. Initially, Sinclair approached Olsen solely with a view to using Timex for the complex flat-screen production. The relatively straightforward ZX81 assembly could have been handled by any one of a number of alternative subcontractors. Desperate to develop the diversification strategy as quickly as possible, Timex persuaded Sinclair to let the Dundee plant handle the microcomputer assembly. By December, 50,000 units were rolling off the production line.

Since it affects later developments in the Sinclair story, it’s worth mentioning that Timex’s diversification programme also included manufacturing contracts with IBM and Nimslo International. Between 1978 and 1981, Nimslo raised around £27m from British institutions for the development of a revolutionary 3D camera. Now, since Fred Olsen was a major shareholder in the company that controlled Nimslo, it made sense, to him at least, that production of the camera should be handled by Timex, Dundee. As it turned out, the Scottish plant failed to come up with the goods. The Nimslo camera was launched in the States in April 1982, with fewer than 60,000 cameras divided among 7000 dealers. Less than half of the production target of 200,000 units had been manufactured, and many of the completed cameras were reportedly substandard. The Nimslo contract was eventually turned over to the Japanese, which, as we shall see, was to have serious consequences for Sinclair’s plans.

In January 1981, Sinclair had managed to put together a £5m investment package for flat-screen development. Sinclair’s choice of Timex, Dundee, for the pre-production plant had turned out to be to the advantage of both companies. Government schemes to develop employment on Tayside facilitated a £2.6m grant to Sinclair Research from the DOI and the Scottish Enterprise Planning Department.