The Perfect Machine (56 page)

The Pasadena scientists and engineers were undaunted by either expertise or experience. Francis Pease and Sinclair Smith had put in their time on large telescopes, and they understood what an observer or an astrophysicist wanted from a big telescope. If they weren’t trained engineers or experienced at solving engineering problems, they had the background in physics that let them understand problems. And because everyone in the group shared that background—the Caltech-trained engineers like Mark Serrurier had strong training in
physics and math—the problems and solutions to the design of the telescope received a thorough grilling.

Sandy McDowell knew whom to see at every big company, knew how to get through to the president of the company, the head of production, or the head of sales. But while he could negotiate a contract with Westinghouse or Babcock & Wilcox and could list specifications for what they were to produce, at bottom he lacked the common background in science that had been so important in the conception and design of the telescope. Men like Rein Kroon or George McCauley, men who could understand and explain problems in the language of physics, were quickly welcomed into the group designing and building the telescope. McDowell remained an outsider.

To McDowell the scientists—Pease, Anderson, Adams, Smith—were unrelenting theorists, pie-in-the-sky dreamers who would go on designing a telescope forever, working out theoretical problems, reinventing the wheel as they conceived solutions to problems that in only slightly different form had been solved elsewhere. When they tried to explain that the telescope was a unique challenge, that no machine had ever been built to that scale and those demands for precision, he would point out that a triple gun turret on a battleship was
almost
as large, that the targeting requirements for a naval gun were only a few orders of magnitude less exacting. Their insistence on exploring, designing, and engineering every step of the project from scratch—doing basic research on subjects as varied as oil bearings, the wind resistance of dome sections, and the chemistry of glass—was for McDowell a sure route to a quagmire of indecision that would never see the telescope built.

McDowell believed he had gotten the scientists off dead center by forcing the end of design and the beginning of construction. By 1937 Babcock & Wilcox was near completion of the complex welded mirror cell, Westinghouse was well along on the tube and yoke, the astrophysics machine shop working on the demanding job of grinding the gears that would set and drive the telescope, and the work on the peak was moving along so well that visitors had begun to come up in droves to admire the buildings and sites. Through it all McDowell seemed always to see problems of management, of breaking up logjams, attending to details, keeping the project moving. A memo ordering the use of spring faucets in the showers and washhouse on the mountain to conserve water would receive the same attention and priority as a discussion of the control system, bearings, corrector lenses, or even the state of the primary mirror of the telescope. For McDowell the telescope was a project. He had done others before and would do others after.

Men who were building a unique machine, an instrument perfect enough to explore the secrets of the universe, didn’t like that attitude.

26
Fine Points

There was no one moment in the optics shop when the big disk was suddenly all right. By January 1937 George McCauley had identified the cause of the fractures as contamination from the refractory bricks in the melting tank and offered his best-guess explanation that the contamination had been introduced by drainage down the sides of the tank as the level of glass in the tank fell. If he was right, the first Pyrex poured into the mold, from the top of the melting tank, should have been pure, and the checks and fractures would begin to disappear as the surface, the material poured last, was ground off.

At the end of January 1937 Amory Houghton and McCauley traveled to Pasadena for another secret meeting with Anderson, McDowell, Mason, Millikan and Weaver from the Rockefeller Foundation. McCauley described the various experimental procedures he had followed and his plans for producing a new disk if one was needed. Houghton thought his offer to produce a disk at Corning’s cost generous; Mason and Millikan thought Corning should at least share the expense. Houghton, secure in McCauley’s assurances that the disk they had already produced would prove satisfactory, said he would discuss the possibility of Corning sharing the expense with his board of trustees.

McCauley’s prediction was as much hope as science, but as the flat grinding approached the final level at the edge of the disk, the checks began to diminish. A few troublesome fractures remained on a cord between two of the pouring points, but the frequency of new fractures decreased, and the gravity of the remaining fractures was less troublesome than the early deep ones. In midsummer 1937, the correspondence about the disk was all still in letters marked PERSONAL, but John Anderson and Max Mason had let their tentative negotiations for a new disk lapse.

Despite the good signs, the initial grinding had gone
much
slower than anyone predicted. The disk had been in the shop for almost a
year and Marcus Brown and his crew still hadn’t begun the real task, grinding a precise concave shape that would transform the glass disk into a telescope mirror.

“How long will it take to finish the mirror?” everyone asked—reporters, visitors, even officials from the Rockefeller Foundation. For those who read or heard about the telescope episodically, their attention piqued only when press conferences, public announcements, magazine features, Robert Benchley’s jibes, or spectacles like the railroad journey of the disk from Corning to Pasadena brought the telescope back to public attention, it seemed as though the project had already gone on for as long as anyone could remember. In 1928 George Hale’s predictions that the telescope might take as long as six or seven years to build had seemed pessimistic. By 1937, almost a decade into the project, the actual shaping work on the mirror hadn’t even begun. Figuring the mirror of the one-hundred-inch telescope had taken almost six years, and that was working with plate glass, a familiar material. After the preliminary grinding of the faces of the Pyrex disk had taken a large crew of men most of a year, and used up five tons of carborundum to remove two and one-half tons of glass, no one on the project needed more evidence to realize that even the most pessimistic estimates of a finishing date were wildly optimistic.

Though it seemed an eternal job that would leave everyone in the optics shop permanently deaf, the surface grinding finally halted in the spring of 1937. There were still visible checks in the disk, but it looked as though they would be ground away when the disk reached a concave shape. The next work, grinding the perimeter of the disk, the edges of the central hole, and reaming out the pockets for the support system, took three months. Special equipment had been built for these tasks, using hollow iron cylinders fed with water and carborundum as the grinding tools.

To visitors the work in the optics shop looked and sounded the same from one day to the next. The disk was on a rotating table in a huge machine. Men hovered over and around it, tending the slowly rotating tools that ground away the glass. But to the optician, there was a great difference between the “mechanical” work of preparing thirty-six pockets, each of precise dimensions and spacing, or trueing the edge of the disk—all work done to tolerances that the best of machine shops can achieve with care—and the ultimate optical work, figuring the surface shape of the disk to tolerances unknown outside the optics lab. The edge grinding and the preparation of the pockets was boring and tedious. Everyone in the shop looked forward to the start of making a mirror.

It was almost the anniversary of the disk’s arrival when Marcus Brown ordered a thorough cleaning of the shop, the sort of search-and-scrub that made the normal daily cleanings seem casual. The turntable of the grinding machine was covered with two layers of one-inch
sponge rubber. To assure even seating of the disk, the sheets of rubber had been tested with a fixed load; only sheets with measured and tested compression within a tight range were used on the table.

Corning had cast a forty-inch Pyrex disk to fill the center hole in the mirror disk during the grinding of the mirror. Trueing the plug to a perfect circle was a relatively simple task, but Anderson had to think awhile to come up with a scheme to lower the 1,400-pound plug into the hole in the disk, keep it there during the grinding and polishing of the mirror, and be sure of removing it without harming the disk. His solution, described regularly to visitors, was a staple story for the optics shop, guaranteed to evoke “Why didn’t I think of that?” smiles.

Anderson designed a wooden lifting clamp that was attached to the top half of the plug, leaving the lower ribbed section, approximately fifteen inches deep, projecting. The overhead crane could then lift the plug and swing it over the disk. To lower the plug into the hole in the main disk, Anderson had Brownie’s workmen place a large cake of ice, tall enough to support the plug, in the hole in the disk. The crane lowered the plug in place onto the ice, the lifting frame was removed, and as the ice slowly melted, the plug slipped into its exact fit in the disk. For optical lab workers, used to watching minuscule progress after a day or a week of polishing a mirror, watching ice melt wasn’t boring. A room full of workmen exhaled all at once as the plug settled into place without harming the mirror. Brownie used plaster of paris and waterproof cement on the seam to hold the center plug in place.

Brownie was almost ready to start shaping the disk. During grinding and polishing the disk would rest on its back, supported by foam rubber cushions on the machined turntable. For testing, the disk would be raised to a vertical position, both to permit the long light path that the test procedure required and because a vertical orientation would more closely simulate the most extreme loads on the disk in use. In the vertical position the disk would require the internal support system suggested years before by Elihu Thomson, and refined in years of sketches by John Anderson, drawings by the Caltech engineers and draftsmen, and model making in the astrophysics machine shop.

The internal supports were amazing machines. Counterweighted levers were engineered so that in any position, the weights would exert the correct forces on the back of the disk to compensate for the sags or other deformations in the overall shape of the disk. The forces required are so small that the fundamental problem in designing the support system was what an engineer calls stiction, the initial friction of a system that is mostly
not
in motion. Bearings and lubricants were tried and rejected. The design grew more complex, until there were dozens of parts in each support. The thirty-six supports would work together, automatically translating the motions of the counterweights
to the precise combination of tiny pressures on the disk that would compensate for its changed position. The blueprints of the devices got so complicated that only a few men in the machine shop understood how to put the devices together. Among those who didn’t have to decipher the drawings, the supports were black art. Rumors started that only one or two men on the Caltech campus understood the supports.

The mirror supports were installed in their pockets in the back of the disk before the shaping of the disk began. Initially the mechanisms were disconnected, lest the delicate bearings and cams be subjected to the motions and pressures of the largest grinding tools. Later, when the disk was ready for testing, the supports were reconnected. The forces exerted by the support system were so small that the only real test of the supports was the optical tests of the mirror. Like the great oil bearings, the corrector lens, the readout devices (servos) to tell where the telescope was pointing, and dozens of other technologies that had been developed and refined for this telescope, the supports were innovations, designed for an instrument so precise that the technologies could ultimately be tested only with the most precise and exacting test of all—the light of a star.

In the summer of 1937 Brownie and his crew began shaping the mirror. Day after day the same small crew of workmen in white cotton surgical suits and canvas shoes worked the big machine. Occasionally they traded jobs. Usually regularity was more important: The same man did the same job each day. One man stood on a scaffold under the machine, greasing the main gears for the turntable to make certain they didn’t gall. Another man continuously washed the edges of the disk with a hose, sloshing away the excess grinding slurry. A man stood by the power switch in case of an emergency. The seventeen-and-one-half-foot-diameter table of the grinding machine turned slowly. The disk was so large that at a rotating speed of one-half turn per minute, the outer edge moved by the grinding tool at only twenty-six feet per minute.

Hour after hour, day after day, week after week, month after month, the disk turned, while the grinding tool above turned in its own serpentine lissajous figures. Occasionally the routine was interrupted so they could change the glass blocks on the face of the tools. Every few months they would change the grinding tool, from the one-third- and half-size tools to the full-size tool, a disk as large as the mirror itself. A sharp-eyed visitor might notice a slight change in the configuration of the machine. Most visitors would watch for a few minutes, amazed that men could work hour after hour, day after day, doing the same job, in the same windowless room, with the same droning machines, and without seeing any progress in their work.

The concave shape Brownie and his men were grinding into the disk would be approximately three and three-quarter inches deep at the center of a two-hundred-inch-diameter circle. It would take months
before the curve was apparent to the naked eye. The men in the room stopped guessing how long it would take to grind and polish the mirror to the approximately one-millionth-of-an-inch precision the final figure would require.