The Perfect Machine

by Ronald Florence

  The test reports on the trial blanks, and the nasty bubbles that had emerged during grinding, argued for thicker sprayed coatings than Ellis and Thomson had planned. The problem was especially acute with the sixty-inch mirrors. The finished telescope would require at least three of these secondary mirrors, with both concave and convex surfaces. To make the raw disks interchangeable, they would need clear quartz coatings thick enough so that the edge could be ground down for the convex mirrors and the center ground down for the concave ones.

  As Ellis stared at his drawings he had an inspired thought. What if they didn’t mold a base at all and built up the entire volume of the disk by spraying? Dispensing with the process to mold the base of the mirror—furnace, tooling, mold, fuel, and personnel—would save money and time. They would no longer have to face-grind the surface of a molded base before commencing the spraying, and there would be no chance of temperature coefficient mismatches between the different layers. Instead of two complex processes to set up, they would have only one. Once the spraying process was in operation, they could spray day and night, as much fused quartz as they needed. It was just a question of keeping the furnace going, and supplying more fuel and ground quartz to the spraying apparatus. They could do mirror after mirror, any size they needed.

  The new plan made sense, but it was an incredible leap into an unproved technology. The largest volume of fused quartz they had sprayed was a layer approximately five-eighths of an inch thick on the surface of a twenty-two-inch disk. Despite every precaution the twenty-two-inch disks had not emerged as usable mirror blanks. The sixty-inch disks would need to be approximately twelve inches thick, an increase in volume of 225 to 1.

  Before he could go ahead, Ellis needed a design for the disk. At the tolerances needed for a large telescope, even seemingly rigid materials like glass and fused quartz are so fluid that the telescope designer must cope with the potential changes in the shape of the mirror disk as the telescope is swung from the horizon to the zenith. The weight of the disk itself, and the forces acting on it from its own mounting, deform the mirror as its orientation changes. The acceptable tolerance is close to zero. Distortions too small to measure mechanically can still be detected optically, and even a minuscule change in the shape of the disk affects the quality of the images.

  Telescopes like the sixty-inch and the one-hundred-inch had been built with massive, solid mirrors, thick enough to maintain their shape as the telescope moved. Solid disks might work for the sixty-inch auxiliary mirrors, but one fundamental concept of the whole mirror project—part of the plan from the time of Hale’s earliest talks with Rose—was that each stage in the work would function as a test bed for the next. The auxiliary mirrors were the design models for the two-hundred-inch mirror. The rule of thumb was that the thickness of a solid disk should be one-sixth the diameter. Following that formula a solid disk two hundred inches in diameter would weigh more than forty tons. From his calculations of fuel consumption and quartz supplies, Ellis knew that a solid two-hundred-inch disk would be impossible to fabricate.

  It would also be too massive to mount in a telescope. If the ratio of diameter to thickness is maintained, the mass of a disk increases with the cube of an increase in diameter. As the mass increases, even a temperature-stable material like fused quartz becomes vulnerable to changes in the ambient temperature. If the air outside the observatory dome was ten or fifteen degrees cooler than the air inside, not at all unusual in the early evening at a mountaintop observatory, when the dome was opened to use the telescope, the surfaces of the mirror would quickly begin cooling to the ambient temperature. The rest of the disk would cool at a slower rate, as the cold was gradually conveyed to the interior of the mirror. Depending on the mass and thickness of the mirror, for hours, perhaps for an entire observing session, there would be differences in the temperature of different portions of the mirror. With one part expanded and another contracted, the mirror would bend and flop, distorting the images it focused on the astronomer’s photographic plates. The greater the mass of the disk the greater these distortions. Even if temperature effects didn’t rule against a solid disk, the sheer weight of a mass of quartz that huge would make it unusable. A mounting to hold it would be prohibitive, and the disk would sag and deform from its own weight as the telescope slewed from position to position.

  For more than a year Anderson, Hale, Thomson, and Ellis had been discussing alternatives to a solid mirror disk. They needed a disk sufficiently rigid to hold its shape, thin enough in each dimension that it would avoid the effects of differential cooling, and with provisions for attaching and supporting the mirror that would compensate for any changes in its shape as the telescope slewed from the vertical to the horizontal.

  Hale, who despite the increasingly frequent attacks from his demons, had an idea for every problem, had already put his own suggestion for supporting the mirror in a memo to Anderson. Hale’s idea was to use a liquid or air support under pressure that would vary with the inclination of the mirror. He urged that the idea be tested with a large disk of thin glass plate. An ingenious suggestion, it was never tested or explored because there was no pumping and valving equipment available sensitive enough to control the local pressure of a cell of water or air with the precision needed to shape the mirror as the mounting moved.*

  As they went through different options, the best idea, it seemed, was the scheme Thomson had proposed to Anderson in the gloomy days a year before, of ribbing the back of the mirror like a waffle. If the ribbing were designed carefully, the ribbed back would mean that even with a two-hundred-inch mirror, no portion of the quartz disk would be more than four or five inches from the surface. The disk would respond to changes in temperature like a much smaller disk, and avoid the nightly problems of a thick slab. With the continuing problems of the mirror of the one-hundred-inch telescope in their minds, the ribbed scheme seemed a splendid solution.

  Thomson carried the idea a step farther by suggesting that the pockets between the ribs could be used for an active mounting system to support the disk. If grooves and ribs were molded into the disk, they could be gently pushed and prodded with levers to compensate for the mirror’s tendency to deform as it moved. Everyone began sketching his own version of the waffled back of the disk. Anderson tried dozens of variations, changing the thickness and layout of the ribs, and substituting or adding a pattern of round pockets sunk into the back of the disk along with or in place of the ribs. While Anderson worked on alternatives for the shape of the ribbed structure, Thomson sketched lever-operated actuators that would use gravity to create compensating pressures on various points of the back of the mirror as the telescope was swung from the horizon to the zenith. He used ball-and-socket joints, balancing arms with counterweights, and gimbels to create devices that could be fitted to each pocket in the back of the mirror. If the systems were designed correctly, the mirror would automatically assume the correct shape no matter which way the telescope was tilted or turned.

  It was an ingenious idea. Thomson, an instinctive inventor who loved mechanical gadgetry, sent pencil sketches of the complex machines to Pasadena. No one had ever built such a device, and it seemed fantastic to assume that a mechanical device could sense and make movements on the order of one millionth of an inch to correct the changes in shape of the mirror, but it was also hard to find fault with the design. Thomson’s reputation for genius made even fantastic ideas seem workable.

  Pease, who had been working on the problems of the one-hundred-inch telescope, calculated that ball-bearings on the edge and bottom support systems for the mirror would reduce the friction of flexure and expansion between the glass and metal surfaces from ±0.1 inch to ±0.001 inch. His calculations, and the use of ball-bearings, added another element to the design. Suddenly, it seemed, they had a workable design for the mirror of the telescope. Like so many good ideas, it had emerged not so much from a deliberate research program as from very bright minds freely exchanging ideas, nourished by serendipity and the
willingness to explore radical concepts. Ellis’s idea of spraying the entire volume of the mirror meshed perfectly with the collective ideas of Ellis, Thomson, and Anderson for a ribbed back and a support system for the disk, and Pease’s suggestion for ball-bearing supports to reduce the friction between the glass and the metal of the mounting.

  Now, all they had to do was spray the quartz mirrors, grind and polish them to nearly perfect optical surfaces, build a telescope around them, and an observatory for the telescope. When things went well, everything seemed easy.

  13

  Orderly Progress

  George McCauley was an orderly man.

  Each workday morning he walked from his home on Fourth Street, up on the hill above the Corning Glass Works, down to his office or to one of the factory units where he had a project under way. He was a hard worker, with a reputation for concentration. Most days he would work without a break all morning. At noon he would walk home for his dinner and a twenty-minute nap, and then head back down to the factories in time for the one o’clock whistle and an afternoon of more hard work. In the evening, after supper, he would bring out a portable drafting table and a T-square, and do design work or calculations at the cleared dining room table where his children were doing their homework.

  McCauley came to Corning via Northwestern University and the University of Wisconsin, where he had gotten a Ph.D. in physics. He had taught briefly at Northwestern, and worked at the Bureau of Standards during the war. His job at the Glass Works was research on special projects. He liked the company and the town. He had grown up on a farm in Missouri, and he enjoyed the quiet, rural beauty of the Chemung Valley. There were dairy farms and orchards in the surrounding countryside. It was a good place to raise a family. George was a warden of the Episcopal Church and a scoutmaster of the local troop.

  Corning had originally been a railroad town. The company that became the Corning Glass Works was recruited from New York City by a commission appointed to seek new business at the end of the nineteenth century, when the railroads were threatening to pull out. Corning soon gave its name to the glass company, which grew large enough to dominate the town. For the most part it was a welcome domination. The Houghton family, majority owners of the Corning Glass Works and chief executive officers, were popular in Corning. They were fair bosses, and Corning built a reputation as a good company.

  Corning had always been a research-oriented company. In the 1920s it developed a series of new glasses, based on borosilicates fabricated at very high temperatures. Under the trade name Pyrex, these new glasses revolutionized house wares. A Pyrex pie plate could go directly from the icebox to the oven, without danger of cracking or exploding. For housewives Pyrex ware was a timesaving revolution. Pyrex pie plates didn’t crack because the borosilicate glass formulation had an extremely low coefficient of expansion—exactly the quality astronomers sought in telescope mirrors. It didn’t take long before astronomers inquired about the new glass.

  As far back as 1922, Corning had received orders from the Mount Wilson labs for small glass disks. In 1929 a query arrived from Yerkes Observatory. George McCauley pointed to the queries and the potential for customers at other observatories and got authorization to cast a twenty-seven-inch-diameter disk with a four-inch-diameter hole in the center for the sales department to use in approaches to observatories and at the annual convention of the AAS.

  It seemed a simple enough project. McCauley had the Corning masons build a mold from insulating brick, using an iron strap around the bricks to hold them in place. The bottom of the mold was clay damper tile. To mold a central hole in the disk he had the technicians use more of the same C-22 refractory brick material, held in place with Hi-Tempite cement to withstand the heat of the molten glass. The mold was finished in May 1929, during a busy period in the old A Factory on the riverbank. McCauley had it set aside against the wall of the factory to await a slack period.

  Through the summer of 1929 the factory was working long shifts. McCauley’s mold became a convenient table for sundry objects, and by August, when he was ready to try casting a mirror, the central core of the mold had broken. A Corning mason did a quick repair, using wet clay and cement. As the mold was filled with molten glass, the moisture in the core launched a cloud of bubbles in the glass. McCauley told the technicians to add more glass to fill the voids. When the mold was full, he had it moved aside, out of the way of the usual traffic in the blowing room, so it could quietly anneal. Glass foundries are huge, noisy places. On the vast factory floor McCauley’s experimental mirror mold was a small project. If this was all it took, McCauley thought, astronomical mirrors seemed an easy sideline for Corning.

  The annealing went well until the temperature inside the mold reached 480°C. At that point, the heat triggered a sprinkler head on the ceiling, part of the factory’s fire protection system, dousing the mold with water. After some laughter at McCauley’s expense, technicians turned off the sprinkler, and the cooling process continued until the disk was cool enough to uncover. The disk emerged from the mold with traces of bubbles in one area from the hasty core repair, and the interrupted annealing was less than successful, leaving strains in the glass, but the disk had survived in one piece. McCauley pronounced the experiment a success and gave the go-ahead to the sales department to try to solicit orders for disks for telescope mirrors. As an added incentive for their presentations to observatories, he suggested that it would be easy to mold a curve into the face of the disks, which would cut down on the initial grinding when the disks were shaped into telescope mirrors.

  McCauley’s preparations came at the right time. George Hale’s article in Harper’s and the publicity surrounding the announcement of the grant for the two-hundred-inch telescope had started a wave of telescope building. The Universities of Michigan and Texas were shopping for telescopes in the eighty-four-to one-hundred-inch range. The Perkins Observatory was already building a sixty-nine-inch telescope. George Ward and Wilbur Foshay at Corning put together cost figures for different-size disks, based on McCauley’s estimates.

  Queries came in, but the Corning salesmen couldn’t land a large order. They could argue that their product was superior, but their prices were high, at least compared to the quotes that the universities had received from glass foundries abroad. A flurry of interdepartmental memorandums went back and forth, questioning whether Corning should make an effort to quote cheaper prices for the big disks. The initial answer was no. On the basis of the special equipment and man-hours the project would need, the small market for telescope mirrors didn’t seem a profitable business in 1929.

  A year later the answer changed. By 1930 the depression had touched almost every sector of the economy. In some industries, like steel in Pittsburgh, more than half the workers were laid off, and companies devised schemes like the “stagger plan” to let men share jobs (which left them ineligible for unemployment compensation). The Houghton family were reluctant to lay off any workers at Corning, no matter how severe the economic situation. The chance to keep a few more workers busy, to strengthen Corning’s role in glass research, and to build more ties to university research departments was enough incentive to aggressively bid for the potential telescope business.

  McCauley went to the Bureau of Standards, where he had worked before coming to Corning, to survey the procedures they had used to cast the mirror of the Perkins telescope, the first large glass disk ever made in the United States. The tank melting and ladling procedures Corning had already developed were as good as the Bureau of Standards procedures. But casting a disk was only half the battle. Annealing the disk, subjecting it to long, slow, controlled cooling to minimize strains in the glass, seemed to be the real challenge. The bureau experiences confirmed what McCauley had learned in his initial trials at Corning: The high temperatures and long duty cycle for annealing would destroy all but the most rugged, commercial-grade heating units.

  McCauley spent more evenings at his oak dining table. To minimize the investment in what
would clearly be a small sideline for Corning, he tried to use commercially available materials and facilities. The molds, furnaces, annealing kilns, cranes, and lifting slings could all be fabricated by Corning craftsmen, who might otherwise be idle. When McCauley demonstrated that the new business would require no large outlays for special equipment or custom-fabricated tools, and that it would provide work for underemployed Corning workers, the Corning bean counters recalculated their prices for disks.

  In March 1931 the new cost estimates paid off. Corning received a firm order for an elliptical auxiliary disk for the Perkins telescope. McCauley was ready. But one order didn’t make a telescope disk business.

  At GE’s West Lynn laboratory, Ellis got ready to spray a sixty-inch quartz disk that would be used for one of the auxiliary mirrors in the telescope. The huge sheet iron building, the custom furnace, built from special ceramics and graphite developed by the Carborundum Company and the National Carbon Company, and the huge, rigid copper bars and regulators to power the electrical furnace for this stage of the process had cost more than $115,000. Ellis justified the expenditure by noting that the building, the special ceramics for the furnace, and the electrical equipment would all be reused for the two-hundred-inch mirror.

  Every time there was a choice in equipment or facilities, Ellis, with the approval of Thomson and the GE management, opted for the more versatile, and invariably more expensive, item, on the grounds that more versatile equipment could be adapted to larger mirrors and would have higher resale value. Instead of inexpensive insulated cables to bring electricity to the furnaces, he used expensive copper bus bars. The building was equipped with a seventy-five-ton gantry crane, far larger than needed to fabricate or move a two-hundred-inch disk. Whenever Hale or Anderson questioned the budgets, Ellis explained that each decision had been made in the interest of salvage value. No doubt his arguments were true, but it was also true that each choice favored continued productive capacity at GE over cost control and expediting the production of the mirrors for the telescope.