Celebrations are in full swing to mark the anniversary of the first computer. That depends, of course, on which computer you mean. Jon Courtenay Grimwood begins our hardware special with a look at the dawn of the digital world
So utterly confusing is their historic order of precedence that if electronic computers were people going in to a formal supper you would never be able to get them in through the door. Come the end of the meal everybody would still be stuck outside arguing about who really came first.

Of course, it all depends on what you mean by "computer". If we leave out the word "electronic" then some people would argue that Charles Babbage's 1833 "difference engine" was the first general-purpose computer; others would say it is nothing but a large calculating machine. A replica of Babbage's second, 1847, engine was constructed in 1991 from the original drawings under the leadership of Doron Swade, the National Science Museum's senior curator. According to Mr Swade, Babbage's drawings survived intact only because there was no attempt to build the second "vastly more elegant, twice as efficient" machine. Drawings for the first engine were damaged and lost in the workshop.

Unfortunately Babbage was hindered by a lack of government funds - some things in science never change - and was unable to complete the construction of his engine. But the machine had a store (memory), a mill (cpu), and a printer - much like a modern PC. Mind you, it was also to be steam-driven, have 50,000 moving parts and be the size of a train!

Historic computers are back in the news - not least because we have reached (or are reaching) the 50th anniversary of many of the machines which lay claim to be the first. That this is being celebrated - not least on Internet sites such as the Virtual Museum of Computing - shows that the industry has reached a sort of maturity. We did not always have this reverence for old computers. You only had to walk by any overflowing, midtown skip in the late 1980s to see cracked monochrome monitors next to discarded 286s, the detritus of an office-based culture moving up a gear.

Most old computers still get dumped, of course, as surely as outdated ranges end up at mail-order houses. But our attitude to obsolete machines is beginning to change. For a start, otherwise ordinary people have begun to collect classic examples.

Take Cornishman Jeremy Rundle, who has more than 30 machines ranging from the once-ubiquitous Sinclair ZX81 to a working example of an Osborne 01, which Mr Rundle cites as the first portable computer. Weighing in at 28lb and operating from two floppy drives, the Osborne cost pounds 1,250 and caused a sensation when it was released in 1981. That may sound recent, but in terms of the exponential advance of computing it is prehistoric.

It is the museums and universities that are heading the drive to preserve our technological heritage, by building working copies of the great valve- driven machines from the early days of computing. The crown for "first general-purpose all-electronic computer" is usually claimed by Eniac, which was built in 1946 at the University of Pennsylvania and which celebrated its 50th anniversary in February. Eniac contained 17,468 vacuum tubes and 500 miles of wiring, weighed 30 tons and was a staggering 18ft high by 80ft long. It also had a then-miraculous speed of several hundred multiplications per minute. However, its program was hard-wired into the processor, which meant changing the program required rewiring the machine by hand.

There were "computers" before Eniac of course, but either these were not fully electronic or they did not use vacuum tubes. Harvard mathematician Howard Aiken completed the best known of these in 1944. Mark 1, as it became known, was designed to create ballistics tables aimed at making US naval artillery more accurate, and relied for its processing power on 3,304 "electro-mechanical relays".

But even before this, in 1941, the German scientist Konrad Zuse had produced the Z3, described by the Berlin Science Museum as "the first fully functional electro-mechanical computer in the world". Zuse, who died late last year, intended the Z3 to be used in the design of military aircraft and missiles. Luckily for the Allies, he was denied sufficient funds to fully develop his machine and the Z3 never achieved its potential.

But if Eniac claims the crown for "first general-purpose all-electronic computer" the title of "first single-purpose all-electronic computer" must go to Britain's best-kept secret: Colossus, the huge cypher-breaking machine inspired by the undisputed father of computing, the English mathematician Alan Turing.

Designed and built at the Post Office Research Laboratories at Dollis Hill in North London to help Bletchley Park decode intercepted messages from the German high command, Colossus was in operation by 1943 and was capable of working at nearly 25,000 characters per second as it compared encrypted German secret military codes against codes that had already been deciphered. It was the ability of Colossus to decode the signals Hitler was sending to his generals that gave Eisenhower and Montgomery vital information prior to D-Day.

Colossus is being recreated under the guidance of Tony Sale, the secretary of the Computer Conservation Society, which is a co-ordinated project between the Computer Society and the National Science Museum.

The rebuild started two years ago with the collection by Mr Sale of all available information about Colossus. It is typical of the way our computer heritage has been treated that one of the first things he discovered was that the original, priceless machine drawings had been burnt in 1960 as part of a clear-out. However, by poring over old photographs and using a 486 running EasyCad software, it proved possible to recreate the drawings in three months. The rebuild of Colossus is now working, albeit slowly, and Mr Sale is still looking for valves to populate all the racks.

Classic computers do not have to be military in origin, as proved by the upcoming 50th birthday celebrations for the Manchester Mark I. Claiming the right to call itself "the first stored-program, all-electronic, digital computer" and regarded by many as the first true computer, "Baby", as it was nicknamed, first ran on 21 June 1948. Commercialised by electronics firm Ferranti as the Mark I, it was this machine's ability to store a program that set it apart from the "computers" that had gone before.

Like Colossus and the Difference Engine before it, Baby is also in the process of being reborn, this time with funding from ICL. A replica is under construction by members of the CCS, and will run its first program on 21 June 1998 as part of the celebrations for Manchester Mark I's 50th birthday.

And then there is Edsac. Billed by supporters as "the world's first practical stored-program computer". It was developed at Cambridge University and first operated on 6 May 1949. There will be a 50th anniversary celebration for that too.

Z3, Mark 1, Colossus, Eniac, Baby, Edsac - they all have a right to claim to be first, in their own way. And all should be celebrated, because 50 years is 50 years and this is our heritage. But who knows if what we now call computers will still be regarded as such in 50 years? There is a basic law that when we forecast the future, we get it wrong. Writing in 1949, a journalist on Popular Mechanics prophesied, "Where a computer like the Eniac is equipped with 18,000 vacuum tubes and weighs 30 tons, computers in the future may have only 1,000 vacuum tubes and weigh only one-and-a-half tons."