The Perfect Machine

by Ronald Florence

  One major question remained for the telescope: the bearings. The entire weight of the moving portions of the telescope, half a million pounds, would rest on the north and south bearings of the yoke. Francis Pease’s drawings still showed huge roller bearings for the north bearing under the horseshoe, but the horsepower needed to move the telescope on roller bearings, and the potential distortion of the rollers from the heavy load, were discouraging. Ball bearings were worse. Even the declination bearings for the telescope tube presented problems. Serrurier’s new design, which lightened the tube, still concentrated the full weight of the tube on the declination bearings.

  Sandy McDowell, who had been brought into the project to supervise construction, was eager to move from design to actual building. His experience in the navy had been with large welded components, and he wanted to have as much of the telescope as possible built from seamless, welded construction. Especially if the welded sections were annealed—a process of heating and controlled cooling, similar to the annealing of a glass disk—which would remove residual strains in the steel, the telescope would emerge a strong, monocoque structure, with no danger of fastenings loosening or corroding.

  Some consultants argued that a welded structure would be too stiff, that riveting the members together would permit a flexibility that would slow down or dampen vibrations in the structure. Welding, by increasing the rigidity, would shorten the period of vibrations, increasing the number of cycles per unit of time. The welding advocates argued that vibrations would only be a problem if they didn’t get the drives and mount right. They had worked long enough on this project that they would get it right. Finally even Pease, the traditionalist among the design staff, came around to the welded construction.

  Welded fabrication had another appeal for the project. It was fashionable. In place of the bolts and rivet heads of the Mount Wilson telescopes, a welded telescope would be sleek, smooth, and streamlined, like the new aerodynamic designs that were just being introduced in automobiles and locomotives. An earlier era had marked its technology with the strength of massive raw-steel construction, producing the Brooklyn and Golden Gate Bridges and the massive riveted and bolted frames of skyscrapers. Welding not only offered the technical advantages of fusing the metal fabrications into large, strong structures but created a look as modern as that of the sleek cars that graced the pages of Life magazine.

  There were few companies with experience of welding large structures, and even fewer engineers who had the experience of designing large structures to take advantage of welded joints. The telescope added the additional requirement that many surfaces of the structures had to be machined to extremely fine tolerances. McDowell concluded that only companies with experience building large hydroelectric turbines and heavy gun turrets combined the two skills. He approached the men he knew: Wylie Wakeman, general manager of Bethlehem Shipbuilding; Homer Ferguson, president of Newport News Shipbuilding; J. F. Metton, president of New York Shipbuilding; and W. W. Smith of Federal Shipbuilding. When they turned him down, McDowell approached the commander of the Mare Island shipyard in San Francisco, asking whether they could do the final machining on the mounts if someone else did the fabrication. None of the companies he approached had any experience with telescopes.

  When the shipbuilders turned him down, he went to companies with experience on hydroelectric turbines and other heavy structures. George Hale had already begun preliminary negotiations with Westinghouse and Babcock & Wilcox, who had both done large welded structures for the Boulder Dam. General Electric had done some heavy welded structures for electrical systems in Russia. Baldwin-Southwark Corp. built locomotives. The American Bridge Company, Warner & Swasey, Budd Steel, and Inland Steel were all interested. Westinghouse sent a vice president, the manager of their huge South Philadelphia plant, and a group of engineers to meet with McDowell. Except for a few government projects, like the Boulder Dam, their business had been slow. The end of 1935 and 1936 seemed an opportune window to keep some idle factory capacity busy.

  While McDowell was negotiating to decide who would build the mounting, the preliminary plans were sent to a blue-ribbon list of independent consulting engineers: S. C. Hollister at Cornell, George E. Beggs at Princeton, W. F. Durand at Stanford University, S. F. Timoshenko of the University of Michigan, John Lessells and George B. Karelitz of Lessell & Karelitz in New York City. The consultants unanimously supported the horseshoe design.

  By the end of 1935, McDowell contracted for the first large fabricated component, the cell that would hold the mirror, which he ordered from Babcock & Wilcox, who had taken on Baldwin-Southwark as a partner on the proposal because a portion of the machining was so large it would require an oversize boring mill at the Baldwin Locomotive Works. On his trips east, McDowell also held more talks with Westinghouse. Their huge South Philadelphia plant built ship turbines. They had lathes, milling machines, and boring mills large enough to machine the biggest components of the telescope, spare capacity in the plant, and an engineering staff with experience on welded structures. Jess Ormondroyd, the manager of the experimental division at Westinghouse, agreed to send a Westinghouse engineer out to California to work on the details of the mounting.

  Rein Kroon was the youngest engineer on the staff at Westinghouse. A Dutchman, he had graduated from the Federal Technical Institute in Zurich, couldn’t find a job in Europe, and was hired as a design school student with Westinghouse by Timoshenko and Lessells, both consultants on the Palomar project. Kroon had been too young to enter the Federal Technical Institute when he first graduated from high school, so he had spent a year working as a volunteer in a Swiss factory. Under the Swiss apprentice system, as soon as a worker announced that he had spent long enough in a job, he was moved to a new position. For a quick learner, a year of apprenticeship meant a wide exposure to machine work.

  Kroon was newly married, with a one-month-old son, when he was told that he would be temporarily assigned to a project in Pasadena, California. He hadn’t heard about the telescope project—the newspapers in Europe were too busy with Hitler, Mussolini, and the abdication of the king of England to chronicle an unbuilt American telescope—and he had to look up Pasadena in an atlas to see where he was going. He carried his son on the train in a basket.

  When Kroon got to Pasadena, McDowell had arranged a welcoming party and found Kroon and his family a bungalow on Los Robles Avenue. The next day Rein Kroon, a tall, lanky man with a soft voice and a remarkably soothing accent for a Dutchman, showed up at the astrophysics laboratory, where he was given an office with a drawing board and a stack of sketches and drawings.

  Kroon had an odd advantage over everyone else on the project. Almost everyone who had worked on the telescope had some experience as an astronomer and ideas of how a telescope should be built. Kroon had never worked on a telescope before and hadn’t even seen a large astronomical telescope. He saw every problem of the project as a challenge in pure engineering. It didn’t matter if the project was a telescope or a jet engine: Engineering was problems to be solved.

  The first problem they handed Kroon was the horseshoe bearing. When he came to Caltech, McDowell had asked around about an idea that had been discussed in the navy—the possibility of using a thin film of oil, under pressure, as a bearing for the weight of the telescope. Pease, pushing for roller bearings, dismissed the idea as impractical. McDowell, who had his doubts about Pease, solicited opinions on the idea from outsiders. J. Emerson, from the navy shipyard at Mare Island, recommended against oil bearings: There were too many variables to control, such as temperature and the viscosity of the oil; it would be too hard to keep out vibrations and tremors; it would require considerable power; and any movement or slippage would wear grooves in the bearing surfaces that would effectively destroy them. As an illustration that the bearings couldn’t work, he suggested the experiment of putting a playing card on a spool; no matter how hard a person blows, he cannot lift the card. Like Pease, he preferred the roller bearings.


  In fact, the only real interest in the oil-bearing idea came from Guenther Froebel, an engineer at Westinghouse, who thought that if the problems could be solved, oil bearings would be simpler and smoother than roller bearings. No one else wanted the problem of trying to design an oil pressure bearing, so it was given to the new man.

  Kroon began by going over the calculations that had already been done. Francis Hodgkinson, the chief engineer at Westinghouse, had done some preliminary work on oil pressure bearings for the heavy rotors in power turbines. The idea of the bearings was that instead of having a metal surface press against metal, the two metal surfaces would be separated by a thin film of oil, pumped into the space between the surfaces. Hodgkinson had calculated the power required to maintain a film in a bearing for the two-hundred-inch telescope, and had come up with six hundred horsepower. Motors that large would need a huge generating plant and would cause enough vibration to shake the entire mountain. Pease looked at Hodgkinson’s figures and gloated.

  Intuitively Kroon guessed that Hodgkinson’s figure was too high. Kroon had a strong background in physics and applied mechanics. The engineering formulas were fresh in his head. He studied Russell Porter’s sketches of a possible bearing and realized that Hodgkinson had calculated the power required as if the entire oil flow were through an orifice. In fact the oil would be spread in a film between the two bearing surfaces. When Kroon recalculated the oil pressure as a film flow, the needed horsepower dropped by a factor of one thousand. With his new calculations, the pumps for oil bearings would need less than one horsepower.

  Kroon showed his figures to other engineers and scientists in the basement of the astrophysics building. He was greeted with skepticism. He offered to demonstrate that his oil bearings would work, and the Engineering Committee approved the expenditure of fifty-eight dollars for an oil pump to test the idea.

  At Westinghouse a test model would require detailed working drawings, approvals of the plans, requisitions for materials and model-makers’ time, layers of authorizations, and an interval to schedule the project in the model making shops. Caltech, even with Sandy McDowell’s efforts to introduce proper procedures, didn’t run that way. The astrophysics machine shop was in the building next door. There were seventy machinists, under the direction of Sherburne, who had been recruited to run the shop from his own machine shop in Pasadena. Sherburne, a first-rate machinist himself, seemed to know the best machine operators and machinists. His shop turned out parts for the Schmidt telescope, new instruments for testing on the Mount Wilson telescopes, and models for the engineers and telescope designers with a minimum of paperwork and bureaucratic hassling.

  When Kroon went over to the machine shop late in the afternoon with his sketches, he was introduced to a machinist and told to explain what he wanted.

  “When do you need it?” the machinist asked.

  “As soon as possible,” Kroon answered. Looking at the huge shop, filled with machines and work in various stages of completion, he assumed that it would be weeks, maybe even months, before he had his model.

  Early the next the morning, the machinist came over with the model Kroon had asked for. After the formality of work at the Federal Institute in Switzerland and the bureaucracy at Westinghouse, Kroon was amazed. The machinist had worked most of the night to finish the model.

  Later that morning Kroon invited the engineers and scientists in the astrophysics building to watch a demonstration. The test model was a three-foot-square steel slab, six inches thick, loaded with lead weights to a total weight of 12,000 pounds, on an inclined plane. Three oil ports in the inclined plane were hooked to a pump and reservoir of oil. The slab was suspended on the equivalent of a hinge on one side so it could be moved over the pads, but the friction of the weight was so great that if the slab was moved from one side of the plane to the other, it remained in place.

  Then Kroon started the pump. Oil pumped through the three orifices and spread into a thin film under the weight. When the pump reached full pressure, Kroon nudged the heavy weight to one side. It swung over, then back and forth. By measuring the amplitude of the swings of the pendulum, Kroon could measure the effectiveness of the oil film as a bearing surface. At the optimum setting, the weight oscillated freely from one side to the other, almost frictionless.

  Even Pease, who had been the most skeptical about oil bearings, came around after Kroon’s demonstration. The clearance between the weight and the inclined plane was only a few thousandths of an inch, but it behaved as Kroon’s calculations predicted: A film of oil a few molecules thick could support hundreds of thousands of pounds of telescope. The coefficient of friction was so low that a fractional horsepower motor would easily move the telescope. Kroon’s experiment was expanded to a large weighted bearing, carrying the estimated five-hundred-ton load of the telescope. Moving it at the normal driving speed required fifty foot-pounds; Kroon calculated that a ball bearing would require thirty thousand foot-pounds of driving force to move the same load.

  Kroon’s final design for the horseshoe bearing used four cavities in each pad for the oil orifices. Limiting the flow through the orifices from 600 to 300 psi (pounds per square inch) provided additional stability. In less than a week, Kroon finished his calculations; his sketches and figures, some on the back of an envelope, went to a draftsman. The bearing problem, which had haunted the telescope since Pease’s earliest designs fifteen years before, was solved.

  Next Kroon tackled the design of the horseshoe. If the telescope was to move smoothly, the bearing surface had to be stiff enough to remain round as the horseshoe tipped from resting on one edge, around through the bottom of the and over to the other side.

  The earliest ideas for the horseshoe were for an open truss work design that would be riveted together in the field, the construction that had been used for the one-hundred-inch telescope. To remain rigid with riveted construction, the horseshoe would have required such heavy plates, and such closely spaced stiffening diaphragms, that the whole horseshoe design was almost abandoned. The strength of a welded construction from solid plates made the horseshoe possible again. The examples of the penstocks and cylinder gates of the huge Boulder Dam showed that massive structures could be welded. Still, the design was tricky. The loads on the horseshoe would put a combination of compression and shear loads on large plates of steel. The completed horseshoe would be some forty-six feet across, far too large to ship as a single unit, so the design also had to permit fabrication in sections small enough to move by rail, ship, and ultimately truck to the mountaintop. The challenge was to design internal stiffening members that would retain the needed stiffness under load and still be arranged in a way that would allow the internal structure of the horseshoe to be welded.

  Kroon knew that it was possible, with enough calculation, to arrive at an analytic solution to the design. But he also remembered from his days as a machinist apprentice how often an analytic design by an engineer had made no provision for the difficulties of manufacture. One afternoon at the cottage on Los Robles, while his three-month-old son played on the floor, Kroon cut out sections of cardboard and glued them together in different configurations. He tested each model with tiny weights, measuring the deflection of the cardboard panels with a ruler. By the end of the afternoon, he had an internal design that met all his requirements. When he calculated the stresses the next week, they worked.

  Kroon found the style of work in California, where an engineer could take rough drawings to the machine shop without review, authorization, or certification, both refreshing and productive. Westinghouse had sent him to California for six months. In the first two months he had solved two of the thorniest design problems. Even Pease was delighted to see more of the design details directed toward Kroon.

  Serrurier’s tube design, with his diagonal trusses, had solved the problem of building a relatively lightweight yet rigid tube for the telescope. Kroon was handed the problem of how to pivot the tube on bearings. Ball bearings could carry the load of
the tube—only a fraction of the load on the horseshoe bearings. But any play in the mounting or strains passed to the bearings would distort the tube, and the careful alignment of the optical system. Theodor von Karmann, from the Aeronautics Department, had been studying the problem of the declination trunions, and had his men calculating the stresses.

  Kroon had never seen a big telescope. He had no idea how tubes were usually pivoted in a telescope. But as soon as he saw the problem, he had an idea. The spokes of a bicycle wheel are stiff in tension and compression, though flexible in bending. A light bicycle wheel remains round even with a heavy rider going over a big bump. A ring of spokes, running from the telescope tube to the bearings, he reasoned, would locate the tube firmly. This problem was too tough to model in cardboard, so he calculated the loads and drew up sketches for the draftsmen. Once again his solution looked too simple. Von Karmann, the famed Caltech professor, looked at the thin spokes and asked, “Did you calculate the buckling loads?”

  Kroon, awed that a world-famous professor was reviewing his sketch, hesitantly pointed out that the tension on the opposite spoke would balance the compression, maintaining the system in alignment. Von Karmann agreed, and yet another solution from the gentle Dutchman found its way into the telescope.

  Kroon’s gentle manner served him well. Though an outsider among a crew of Caltech astronomers and engineers, he was accepted as one of the team, invited on the camping trips in the desert led by Russell Porter, where the nighttime naked-eye star observations were interrupted by coyote howls. Kroon was in Pasadena when the mirror arrived, and like the others he felt the sudden glare of the publicity spotlights on the project he hadn’t even heard of a year before.

  His last project was the south bearing of the telescope, which would share the load of the entire fork with the great horseshoe bearing. He spent the Easter weekend “monkeying around at home” in the little Los Robles cottage, trying to think of a design that would create no forces to disturb structure. When he got stuck in his monkeying, he would practice music or play with his infant son. Finally, what emerged was a ball on oil bearing pads. The ball design would work under the enormous thrust loads of the south bearing, and could be powered by the same pump system that served the pads for the great horseshoe.