The replica project includes, from left: Lee Harker, Delwyn Bluhm, Jerry Musselman, John Erickson, David Birlingmair, Gary Slelege and John Gustafson.
by Thomas R. O'Donnell
The Des Moines Register
Even with their modern methods and years of computer savvy, the team from the Ames Laboratory at Iowa State University has found itself almost as stumped as Atanasoff was when he aimlessly drove to Illinois in 1937. The ideas that would make the computer come together hit him as he sat at an Illinois bar late that night.
The modern scientists are building a working replica of the Atanasoff-Berry Computer, the Rube Goldberg-like device built at ISU that proved to be the grandfather of computers now found in many homes and offices. The job has required them to be part engineer, part gumshoe and part rummager.
The replica project includes, from left: Lee Harker, Delwyn Bluhm, Jerry Musselman, John Erickson, David Birlingmair, Gary Slelege and John Gustafson.
The original computer was scrapped long before anyone realized its significance. Most of the original plans and notes went out with the trash, too, leaving the team to rely mostly on old photos and recollections as they try to figure out how it worked.
"There are a lot of places where we're still making the best guess," as to the computer's workings, said team member Harold Skank, an Ames Lab electronics engineer.
It's been just as tough to find the outmoded parts the computer used. The scientists have scrounged through old warehouses and combed through catalogs to find the pieces of the giant jigsaw puzzle.
Even though they have the advantage of advanced technology, "We're going to have to spend the same number of hours (Atanasoff) spent," said Delwyn Bluhm, manager of research and development engineering at the lab and leader of the replica project.
If authenticity weren't a problem, the researchers could easily make the machine work with a microchip here and a transistor there. "We had to decide, how accurate are we going to be?" said Bluhm. "Do we make it look authentic on the outside and cheat on the inside? . . . We decided no."
The whole trip has been so hairy that team member John Erickson "feels we owe him a round trip to a bar in Illinois," Skank joked.
Atanasoff was a professor of physics and mathematics at then-Iowa State College in the 1930s when he began looking for a way to speed the solutions of linear simultaneous algebraic equations. Calculating machines available then didn't have the capacity the problems needed.
Atanasoff realized the machines could be simplified by converting the numbers from a base-10 number system to a base-2 system--the binary system used in computers today. He and Berry, a graduate assistant, got $650 from the college research council to build a machine Atanasoff said would be "the most powerful computing machine in existence."
They succeeded but never filed a patent application on the device. Atanasoff went off to war, Berry graduated, and the ABC's parts eventually were scattered. Little of their plans and notes remain, either.
"There really wasn't a lot that was very helpful to us," Bluhm said.
The team does have some schematic drawings of some of the computer's modules - units that each had a different job - but there was nothing on how those units were hooked together.
"Some of the wiring, we're going to have to guess where it was," said Gary Sleege, an electronics design engineer who is helping put the parts together. But most of the wiring based on conjecture will be hidden behind panels and won't spoil the authentic look of the replica.
The team also started looking for people who had seen the computer or worked on it. They got calls from "people who had heard of somebody who had heard of somebody who had heard of somebody," said team member John Gustafson, an Ames Lab computational scientist. Sometimes, they found "people who were telling us all kinds of things that couldn't possibly be true," Gustafson said.
But they also found people like Robert Mather, a retired physicist in Oakland, Calif., who wired the original ABC as a graduate student t Iowa State. Mather was proud of the neat way he weaved the wires through the computer's frame and took some of the most detailed photos of the ABC that the team has found.
Mather also helped solve a key puzzle - where the computer spit out the answer to the problem it was solving. "You can imagine, we're started on this project and we're trucking down the road on it, and we don't know where the output is," Sleege said.
To see the answer, the computer operator had to stand close to the machine and look down at an odometer-like device inconspicuously attached to an axle, Mather said.
Jean Berry, wife of the late Clifford Berry and Atanasoff's secretary, solved another mystery.
The ABC didn't have a core memory where the computer could put information it produced - 1,500 bits per second - while it worked equations. Instead, it was stored by electrodes that would burn minute holes into pieces of paper. The holes would later be read by another device.
The problem was finding the right paper. The electrodes set some on fire; other kinds were too heavy for the job.
The team asked Jean Berry, who now lives in Alaska, what kind of paper her husband used. "She said right off the top of her head, 'Strathmore No. 2, available at Vroman's Bookstore in Pasadena, California,'" Gustafson said. Unfortunately, Strathmore no longer makes the paper.
Many of the records the team is using were found by two members who traveled to Atanasoff's Maryland home in 1994. Atanasoff, who died earlier this year, was too ill to talk with them, but the researchers returned with prized purchase orders for parts he used to build the machine.
The team has bought dozens of old vacuum tubes, some still in the original World War II military supply boxes, from antique radio supply houses and other sources.
The Chicago company that Atanasoff and Berry bought gears and bearings from is still in business, still makes the parts - and even uses the same part numbers.
But the researchers sometimes found parts by scrounging through old warehouses. Erickson, on a trip to Minneapolis, stumbled into an electrical surplus store and asked for a part used to connect some of the wiring.
The owner was sure he had some - but he didn't want to go into the unair-conditioned warehouse during the summer. He tracked them down a few weeks later.
The team even has found a supplier of wire clad in authentic cotton and rubber insulation, rather than today's plastic insulation.
But they're still in search of the holy grail: an IBM card puncher. There were thousands made and thousands used around the country, but the team can't seem to find one now.
Despite the road blocks, the team believes it's on track to finish the replica by September 1996. They plan to fire it up at least once to solve a problem.
The project has been an education, team members said, and the more they learn, the more they appreciate the genius of Atanasoff and Berry. The ABC, after all, incorporated ideas that are now a basic part of modern computers, like regenerative memory, parallel processing and modular design.
"We never cease to be amazed with how clever and creative they were, Bluhm said.
ISU researchers working on a replica of the Atanasoff-Berry Computer are in need of parts, particularly an IBM model 0010 manual card punch. Anyone with information about the computer or vintage parts that may have been used in it should call John Gustafson.