Sinclair Research

The Cambridge company that built the ZX line, and could never ship on time

Founded in Cambridge in 1979, Sinclair Research built the ZX80, ZX81 and ZX Spectrum, and cut the price of a home computer far enough to create the British bedroom-coder generation. It also shipped late, shipped broken, and took £5 million in QL orders it could not fill. Sir Clive Sinclair handed the whole business to Alan Sugar at a hastily called press conference on 7 April 1986, for £5 million.

sinclair-zx-spectrumZX81hardwarepioneersbritishcambridge1979–1986

Sinclair Research was the Cambridge-based computer company founded by Sir Clive Sinclair in 1979, out of the wreckage of his second company, Sinclair Radionics, and its immediate successor Science of Cambridge. Over the next seven years it produced the three machines that defined British home computing: the ZX80 (1980, £99.95), the ZX81 (1981, £49.95 kit / £69.95 built), and the ZX Spectrum (1982, £125 / £175). These successive cost-reductions in 8-bit home computing brought programming within reach of British families and seeded the bedroom-coder generation that would go on to dominate UK software development through the 1980s and beyond.

The company’s success was the ZX Spectrum, which sold over five million units across its commercial lifetime and was the platform for thousands of British game releases. The company’s stumble was the Sinclair QL (1984), a 68008-based business computer announced before it worked and delivered four months late in a state its own trade press called bug-ridden. On 7 April 1986 Sinclair sold the computer business to Alan Sugar’s Amstrad for £5 million. Amstrad continued the Spectrum line with the +2, +3 and +2A; Clive Sinclair kept the name for other ventures.

Before Sinclair Research

Sinclair Research was Clive Sinclair’s third attempt at a high-technology company, and the entry point to the story is the failure of the second.

Sinclair Radionics, formed in 1961, made amplifiers, pocket televisions and then calculators. Chris Curry joined in 1966; Nigel Searle in 1973. The calculator market collapsed in 1977, and before that the Black Watch — a digital wristwatch — had cost the company £355,000. The National Enterprise Board took a 43% stake for £650,000 in 1976 and injected a further £19 million the following year. Radionics also designed a microcomputer that it sold on to Newbury Laboratories; it surfaced two years later as the NewBrain.

In 1979 Sinclair split from the NEB and set up Science of Cambridge with Chris Curry, which sold the MK14 kit micro. Curry then left, and with Hermann Hauser founded Acorn — the company that would win the schools with the BBC Micro while Sinclair won the homes. Science of Cambridge was renamed twice, first to Sinclair Computers and then to Sinclair Research.

⚠ Acorn is sometimes described as founded by Sinclair Research alumni. The sequence runs the other way: Curry left the predecessor company, and Acorn was already trading before the Sinclair Research name existed.

Five names, one company

The period accounts disagree about when Science of Cambridge was set up — Sinclair User puts it at 1979, CRASH at the tail of the Radionics collapse — and the register settles it by showing that the question is malformed. Science of Cambridge was not set up at all. It was a name put on a company that already existed.

Company 01135105 was incorporated on 19 September 1973 as Westminster Mail Order Limited, and carried five names before it was dissolved:

Registered as	From
Westminster Mail Order Limited	1973 (incorporated 19 September)
Sinclair Instruments Limited	1976
Science of Cambridge Limited	1977
Sinclair Computers Limited	1980
Sinclair Research Limited	1981

So the company that sold the ZX Spectrum was six years old before it was called Sinclair Research, and was first registered to sell things by post. The sequence this entry already described — Science of Cambridge renamed to Sinclair Computers and then to Sinclair Research — is exactly right, and now has dates against it.

⚠ The register gives 31 December for every one of those changes. That is a placeholder used across records of this age rather than four coincidences, so the years can be relied on and the days cannot. Only the incorporation date is a real date.

It also outlived the machines by four decades. The computer business went to Amstrad in 1986; company 01135105 was not dissolved until 14 October 2025.

The cost-reduction playbook

Sinclair Research’s signature was extracting maximum capability from minimum-cost components. The repeated tricks:

Membrane / rubber keyboards instead of mechanical typewriter keys. Cheap to manufacture, divisive in feel — the Spectrum’s rubber keys became iconic but were widely criticised at launch.
Custom ULA chips consolidating multiple functions onto one gate array. The Spectrum’s ULA combined video generation, memory timing, keyboard scanning, and tape I/O onto a single Ferranti-manufactured chip, dramatically reducing component count.
TV-as-display assumption — output via RF modulator to the household television; no dedicated monitor required, no dedicated monitor cost.
Cassette storage rather than disk — every household had a tape recorder; no expensive floppy drive required.
Direct mail order rather than retail — Sinclair Research bypassed retail margins for much of the company’s early life, selling through magazine advertisements. The coupons in Computer & Video Games from November 1981 give the address as 6 Kings Parade, Cambridge, CB2 1SN, FREEPOST, no stamp needed. That model is inseparable from what went wrong next: mail order takes the money before it ships the machine.

The result: each successive machine was cheaper than its predecessor in real terms, and each captured a new segment of the British (and European) consumer market.

The other pattern: late, quirky, returned

The cost-reduction story is the one usually told. Running alongside it, from the Spectrum onward, is a second pattern that the trade press named at the time.

Sinclair User, reviewing the company’s history in June 1986, put it plainly about the Spectrum’s own launch: mail order only, “a disastrous operation which left early purchasers waiting two months for delivery. The early machines were operationally quirky and many of them were returned by customers. The pattern of late deliveries and unreliability had begun to emerge.” The same paragraph concedes what kept the company alive anyway — “notwithstanding, the machine was outstanding value”.

The peripherals fared worse.

Microdrives (£49.95, 1983), one of Sinclair’s pet projects, were “badly savaged by the press”: they initially took longer to load than cassette tape, the tape loops stretched and corrupted what was stored on them, and the drives were unreliable.
ZX Interface 2, with its ROM cartridge slot and two joystick ports, was withdrawn within two years, and only a handful of ROM games were ever produced for it.

The QL

The Sinclair QL is usually summarised as an ambitious 68008 machine let down by Microdrives. The contemporary account is much harder than that, and it is a story about a launch date rather than a design. What follows is what the QL did to the company; the machine itself has its own entry.

Sinclair announced the QL in January 1984 at £399. Sinclair User: “in a desperate hurry to beat the launch of Apple’s Macintosh, the QL was rushed out in an only partly finished form.”

What followed, in order:

The machine shown to journalists at the launch “was hardly a QL at all”. The operating system on display, written by GST, was replaced soon after by a Sinclair-commissioned system called QDOS — which Sinclair User reckoned inferior to the one it displaced.
Machines arrived four months after the launch, bug-ridden, and with — the detail that says everything — “half the operating system left hanging out of the back” on an external board.
By May, Sinclair Research was holding £5 million against 13,000 orders, of which only a few hundred had been delivered. The company offered apologetic incentives, a free RS-232 lead among them.
A reliable working QL did not appear until late summer 1984, by which point Sinclair had been criticised by the Advertising Standards Authority and, in Sinclair User’s phrase, “hounded by customers”.

The figure worth sitting with is the third one. £5 million taken for machines not yet built is, to the pound, what Sinclair Research was sold for two years later.

1985: the year it came apart

January. The home-computer market nose-dived. Sinclair Research, geared for growth, “became desperate for money, just when nobody wanted to re-order”. The planned stock-market flotation was cancelled — a year earlier, a 10% float had valued the company at over £130 million.
January. The C5 launched and was savaged. It “did nothing to bolster Sir Clive’s reputation”; Sinclair User judged that it “branded him in the media as a kind of ‘mad boffin’”. Sinclair Vehicles scaled down and stopped.
Summer. Robert Maxwell announced a takeover of Sinclair Research, widely reported at the time as a rescue — Acorn User was joking about it in print, Sinclair having simultaneously turned up as a cartoon character in Maxwell’s own Daily Mirror. By autumn Maxwell’s board had rejected the deal and Sinclair was on his own again.
October. A financial rescue package satisfied the creditors, who by then included Thorn EMI, Timex, Barclays Bank and AIB.
Christmas. Most of the debt was repaid through a £10 million bulk purchase of Spectrum Pluses by Dixons — the company that had spent its early life avoiding retail was saved by a single retailer’s order.
February 1986. At the launch of the 128K Spectrum, Sinclair predicted his company would again be “the jewel in the British Computing crown”.

Less than eight weeks later he sold it.

Cambridge, and the people who left

Sinclair Research was based at 6 Kings Parade, Cambridge — central, and university-adjacent — and drew on the city’s electronics talent. Industrial designer Rick Dickinson shaped the visual identity of the whole ZX line: the wedge of the ZX81, the Spectrum’s rubber keys and rainbow flash, the QL’s severe black. The software came largely from outside: Sinclair BASIC on the ZX80 and ZX81, and then on the Spectrum, was contracted out.

The Spectrum’s own designers did not stay. Richard Altwasser and Steven Vickers left in 1982 to found Jupiter Cantab, and their advertising said so in as many words: “leading computer designers Richard Altwasser and Steven Vickers… after playing the major role in creating the ZX Spectrum they formed Jupiter Cantab to develop their latest brainchild the Jupiter Ace”.

The Ace shipped at £89.95 by mail order, with Forth rather than BASIC in ROM — a genuinely better idea for a small machine, and, as ACE magazine noted looking back, “this proved to be its downfall”. Jupiter Cantab could not fund the promised add-ons; Altwasser had left the company by 1983. The Spectrum’s architects built a rival to it and the rival died first.

ACE filed Jupiter Cantab among the Cambridge companies as “the ill-fated splinter group from Sinclair Research”, which is the right description of what it was and a fair description of the risk of working there.

7 April 1986

CRASH’s John Minson went to the press conference and caught the tone in a sentence: “If it had happened a week earlier we’d all have thought it was an April Fool, but at the hastily called press conference at noon on 7th April there was no laughter — just the smiling face of Amstrad’s Alan Sugar and the slightly more restrained features of Sir Clive Sinclair.”

£5 million in cash, for the rights to market and manufacture the Sinclair computers, the Sinclair name, and the existing stock. What Amstrad got with it:

Control of the Spectrum 48, Plus and 128, and first option on Pandora, the portable then in development.
A warehouse. Sinclair User reported that Amstrad shifted over 20,000 rubber-key 16K and 48K Spectrums, along with most of the remaining QL stock, to an overseas distributor in a deal “worth more than £2 million” — recovering a good share of the purchase price from inventory alone.
The QL, which it did not want. Amstrad announced within weeks that it would not pursue the machine, and tried to raise a quarter of a million pounds by selling the remaining 19,000 units and throwing in the manufacturing rights free. Interested parties included Timex and the QL user group Quanta; none of them came up with the cash.
Loki, a super-Spectrum “well advanced at the time of the Amstrad take-over”, named for the Norse god. It was never built.

For scale: Amstrad’s profits for the last six months of 1985 came in above £27 million. The company that defined British home computing cost its buyer under two months’ earnings.

The gap between the two men was not only financial. Your Sinclair noted the sight of “the man who did more than anyone to fulfil his vision of a computer in every home, sitting next to Amstrad boss Alan Sugar, who’s called the Spectrum a pregnant calculator”. Amstrad’s stated plans were to move Sinclair machines firmly into entertainment, keep them software-compatible — and, in Your Sinclair’s summary, to “study why Sinclair products have gained a reputation for poor quality”.

Legacy

The ZX Spectrum is Sinclair Research’s most-cited legacy — still commanding over 40 per cent of British machines sold as late as 1986, on Sinclair User’s figures. The cultural effect is harder to count but larger. The British games industry of the decade was substantially seeded by Spectrum-era programmers, and a great many British developers of the 1990s and 2000s started on Sinclair hardware.

Sinclair User’s own summary, written eight weeks after the sale, is the version worth keeping, because it makes the argument in terms of what the machines cost rather than what they did: “Without his sub-£100 micros the Commodore 64 might still be £350 and Amstrad certainly wouldn’t ever have become involved in home computers.”

That is the claim this project cares about. The Spectrum’s importance is not that it was good — plenty of contemporaries were better made, and Sinclair’s own trade press said so at length. It is that it was cheap enough to be in the house, and that being in the house is what produced a generation who could program.

The Sinclair Research name effectively ended with the sale. Clive Sinclair kept it for other ventures — pocket televisions, the Zike electric bicycle — and went on inventing until his death in 2021. The C5 is often filed with those, but it belongs to the story above: it launched in January 1985, while Sinclair Research was still trading, and the damage it did to his reputation was part of what made the following year survivable only on someone else’s terms.