In early 1978, Wayne Ratliff sat in his Los Angeles apartment before a personal computer he had assembled from a kit. The machine, an ISI 8080 with toggle switches on its front panel, had cost him about $6,000, more than two months of his salary as a contract engineer. His goal was modest: to sort football statistics and beat his colleagues in the Friday office pool. The program he wrote in Intel 8080 assembly language he called Vulcan.
He did not know it then, but Vulcan would become dBase, the most widely used database software ever written for a personal computer, the foundation of a $300 million company called Ashton-Tate, and, eventually, the subject of a federal copyright case proving the company never owned what it sold. By day Ratliff worked for Martin Marietta under contract to the Jet Propulsion Laboratory in Pasadena. Two years earlier he had helped write the data management software that supported the Viking lander on Mars. JPL was also the birthplace of the software that would shape his project.
In 1971, two JPL engineers, Fred Thompson and Jack Hatfield, built a mainframe database system called JLDIS, the Jet Propulsion Laboratory Display Information System. Written in Fortran and running on a UNIVAC 1108, it allowed people who were not programmers to create, query, and manipulate structured data using commands typed in something close to plain English. In 1974, another JPL engineer, Jeb Long, added a programming language to it. Because it was built by government employees on government time, JLDIS was public domain software.
Ratliff knew that system well. When he sat down in January 1978 to write a database for his personal computer, he modeled Vulcan on Jeb Long’s version of JLDIS, translating its concepts from a mainframe into assembly language for a microcomputer running CP/M, the dominant operating system of the pre-IBM PC era. The result was fast, flexible, and built around a dot prompt: a period and blinking cursor at the bottom of a dark screen. Type a command, and the machine sorted, searched, indexed, or reported on your data.
No menus, no graphics. Ratliff tried to sell Vulcan himself. An October 1979 advertisement in Byte magazine listed it at $50. Almost nobody bought it.
A later ad priced it at $490, but distribution was nonexistent. By late 1980, Vulcan was dying quietly in the classified sections of hobbyist magazines. Then George Tate called. Tate was a British-born entrepreneur in Southern California who had already co-founded two small software distribution companies with Hal Lashley.
He was not a programmer. He was a marketer. When he saw Vulcan, he saw not a hobby project but a commercial product for businesses. His first piece of advice was blunt: raise the price.
At $50 the program looked disposable. At $700 it looked professional. The same software at fourteen times the price sold more copies. He was right.
In August 1980, Tate and Lashley incorporated their company in Torrance, California. An advertising executive suggested combining Tate’s surname with a word that sounded established and British: Ashton. There was no person named Ashton. No partner, no investor, no silent backer.
The name was a fabrication chosen to sound like old money. The product needed a new name too. Hal Pawluk, who handled marketing, renamed Vulcan to dBase and then made a decision that has confused computer historians ever since: he called the first commercial release dBase II. There had never been a dBase I.
The Roman numeral was a marketing trick, designed to make buyers believe the product had already been through a full version cycle. In a market where software often crashed and lost data, the illusion of a proven second edition was worth more than any feature. It worked. dBase II launched on CP/M at $700.
In its first fiscal year, ending January 1982, Ashton-Tate recorded $3. 7 million in revenue. Then, in August 1981, IBM released the IBM PC, and the landscape shifted overnight. CP/M became the platform of yesterday.
The IBM PC ran MS-DOS, and Ashton-Tate was among the first to port its product. dBase II was available for the new machine almost from the beginning. Tate understood early that the IBM PC was a corporate seal of approval. Every Fortune 500 purchasing department now had permission to buy personal computers, and every one of those machines needed software to manage data.
dBase II was waiting. Revenue climbed from $3. 7 million to $35 million by 1983. By early 1984, Ashton-Tate had sold more than 150,000 copies and controlled roughly 70 percent of the microcomputer database market.
Then, on August 10, 1984, George Tate died of a heart attack. He was 40 years old. The founder was gone, and the company suddenly lacked the man who had built it. David Cole, hired as chairman and CEO in 1981, resigned a few months later.
Hal Lashley departed by January 1986. The company found new leadership in Ed Esber, a Harvard MBA from Cleveland who had previously managed worldwide sales and marketing for VisiCorp. He arrived in late 1984 and began an aggressive expansion. He acquired Multimate, a word processor popular with government agencies and law firms.
He acquired Forefront Corporation and its integrated office suite, Framework. Under Esber, revenue grew 600 percent, from $40 million to more than $318 million. He was also the man who would preside over the decisions that destroyed the company. The central engineering decision of this period came from Wayne Ratliff, who formally joined Ashton-Tate in 1982 as vice president of new technology.
His first major assignment was dBase III. He rewrote the program from scratch in C. dBase II had been written in assembly language, fast but brutal to maintain and chained to specific processors. C was portable.
A program written in C could be compiled for different machines and operating systems. The rewrite meant dBase III could run on the IBM PC, on Unix workstations, on VAX/VMS systems, on whatever came next. Ratliff was not just updating a product. He was building a platform.
dBase III launched in May 1984, months before George Tate’s death. But the version that conquered the American office was dBase III Plus, shipped in December 1985 at $695. It added pull-down menus for users who did not know the command language, an application generator that let non-programmers build functional database applications by answering questions, and support for local area networking, allowing multiple users to share a database simultaneously. It remained fully compatible with the dBase command language that tens of thousands of developers already knew.
By 1985, dBase III was one-third of what the industry called the application trio: Lotus 1-2-3 for spreadsheets, WordPerfect for word processing, and dBase for databases. PC Magazine described the DBF file format, the way dBase stored its tables, as a de facto standard. Other programs adopted it. Ashton-Tate was routinely named alongside Lotus and Microsoft as one of the three largest personal computer software companies in America.
But buried in the architecture was the seed of what would go wrong. The dBase command language was so widely known and documented that other companies began building products that spoke it. The language was not patented, and the conceptual foundation traced back through Vulcan to JLDIS, public domain software built by government employees. Ashton-Tate was sitting on a gold mine whose mineral rights belonged to the American taxpayer.
In 1985, with revenue doubling every year, nobody in Torrance was asking that question. Under Esber, the company pursued acquisitions. None delivered what they were supposed to. Multimate never seriously challenged WordPerfect.
Framework never threatened Lotus. The money and attention spent on these secondary products were not spent on dBase, and dBase was the only product that mattered. In 1987, 63 percent of the $318 million in revenue came from dBase III Plus alone. Meanwhile, the clones arrived.
In May 1985, Nantucket released Clipper, a compiler that took dBase III source code and turned it into standalone executable programs that ran without dBase installed. Ashton-Tate’s own product could not do that. Fox Software released FoxBase, a dBase-compatible database that was faster and cheaper. Both ran code written in the dBase language.
Both were gaining customers. Ashton-Tate’s response was to sue them. Esber publicly confronted the clone makers, asserting that the dBase command language itself was copyrighted. The developer community reacted with hostility.
At one industry conference, the audience chanted “Innovation, not litigation” at the stage. The legal strategy failed to recognize the one thing that might have saved the company: shipping a product so good the clones became irrelevant. That product was supposed to be dBase IV. In October 1988, Ashton-Tate shipped dBase IV.
Seventy-five developers had worked on it. The codebase ran to 450,000 lines. The feature list was enormous: SQL support, a built-in compiler promised for years, a redesigned report generator, a new user interface, transaction processing, and full backward compatibility. On paper, it was the product that would silence the clones.
It was not ready, and shipping it anyway was the decision from which Ashton-Tate never recovered. Complaints began almost immediately. The software was painfully slow. It consumed enormous amounts of memory.
The SQL implementation was incomplete. The promised compiler was not included. Users documented more than 100 bugs in the first months; the company acknowledged 44. By August 1989, Ashton-Tate issued a public statement admitting dBase IV had been shipped prematurely.
The company absorbed more than $60 million in returned inventory. The market did not wait for the fix. Developers looked at the broken product, looked at Clipper and FoxPro, and made the rational calculation. Both worked as advertised.
The clones Ashton-Tate had threatened with lawsuits became the refuge of its own most loyal customers. In 1988, Ashton-Tate held 63 percent of the PC database market. By 1989, that figure had fallen to 43 percent. In August 1989, the company laid off 250 employees.
When the corrected version finally shipped in July 1990, nearly two years after the botched original, the damage was beyond repair. The copyright lawsuit against Fox Software ran throughout this period. It rested on the premise that dBase was original work. It was not.
On December 11, 1990, United States District Judge Terrence Hatter Jr. ruled against Ashton-Tate and invalidated the dBase copyright entirely. The legal basis was the doctrine of unclean hands. Ashton-Tate had failed to disclose to the Copyright Office that dBase was derived from Vulcan, which was itself derived from JLDIS, a public domain database system built by government employees at the Jet Propulsion Laboratory.
The empire that Ashton-Tate had spent a decade building was founded on code that belonged to everyone. Esber had already stepped down as chairman in May 1990, forced out by the board. During his final years he had pursued merger discussions with Lotus, Microsoft, Informix, Computer Associates, and Symantec; the board had rejected every one. By the time he departed, the options had narrowed to a single buyer.
On July 10, 1991, Borland International announced it would acquire Ashton-Tate for $439 million in stock. The deal closed in October. The Justice Department required Borland to cease asserting copyright claims to the dBase command language and to allow other companies to use the Xbase language freely for ten years. The crown jewel of the acquisition was, by federal consent decree, open territory.
The acquisition nearly destroyed Borland. In 1992, Microsoft launched Access, a graphical database built for Windows and bundled into Microsoft Office. In the same year, Microsoft purchased Fox Software and its FoxPro product for a reported $173 million. The twin moves were devastating.
Borland had paid $439 million for a DOS database with a shattered reputation at the precise moment the industry was moving to Windows. The Ashton-Tate name disappeared. The headquarters on Hamilton Avenue in Torrance was vacated. Borland tried to bring dBase to Windows, releasing dBase 5.
0 in August 1994, almost two years after Windows 3. 1 had become the dominant operating system. It arrived too late to matter. In 1999, Borland sold the dBase Windows product rights to a small company called Ksoft, which formed a subsidiary called dBase Inc.
In 2009, Borland, the company that had paid $439 million for Ashton-Tate, was itself acquired by Micro Focus for $75 million. Something survived. dBase Inc. still sells a Windows development environment called dBase Plus that can open the DBF files created in 1985.
The DBF file format has had a stranger and more durable afterlife: when Esri, the company that dominates geographic information systems, designed the shapefile format for its ArcInfo mapping software, it chose DBF for attribute tables. Every map layer in every GIS system used by city planners, environmental scientists, and transportation engineers stores its data in a format that originated in dBase III Plus. The Xbase language family survives in open-source projects like Harbour and in XSharp, which brings it into the Microsoft . NET ecosystem.
Wayne Ratliff is retired. He sails, studies mathematics, and designs computer systems for competitive sailboat racing. In a 1984 interview, he confessed that the success of dBase had left him so busy he had only found time to watch two or three football games that year. The office pool that started everything had become the one thing he no longer had time for.
Ashton-Tate did not fail because dBase was a bad product. It failed because it shipped a product it knew was not ready, sued the developers keeping its language alive, and tried to claim ownership of code it never wrote. The company that ruled PC databases for a decade did not lose to a better technology.
It lost to its own decisions.