The Perfect Machine

by Ronald Florence

  McCauley drew up sketches for an annealing oven and turned them over to George Ward, an engineer who had worked on furnace designs. Ward produced drawings and material lists for the Corning masons, millwrights, and electricians. The first oven, big enough to handle a thirty-six-inch disk, would be used to anneal the twenty-six-inch solid disk and a thirty-inch ribbed disk, the first ones on the schedule for the Observatory Council.

  By March 1932 everything was ready except the annealing oven. George Crown, the favorite Corning mason for lab work, had built molds for the first two disks, shaping the nineteen separate cores that would create the ribbed structure from the reliable C-25 insulating brick, and cementing them to the molds with the Hi-Tempite cement that had been used on earlier molds. Although the annealing oven wasn’t ready, McCauley decided to go ahead by casting the disks, then cooling the blanks slowly under a layer of Sil-o-Cel powder insulation spread over the exterior of the hot molds. The cold disks could later be reheated and annealed when the ovens were ready.

  Except for some bouts of stage fright among the ladlers, the first production casting, of the twenty-six-inch solid disk, went well. The first try at a ribbed disk was less successful. When Woods and his helpers ladled glass into the mold, small bubbles appeared in the spaces between the cores, and the cores slowly loosened and bobbed to the surface of the mold. Tests had showed that the cores should stay in place, but there was no denying the evidence. That evening McCauley went back to his drafting board at the oak dining table where Anne, Jim, and George Jr. did their homework.

  For the next try McCauley had the mold makers use cemented dowels to hold the cores in the mold. The crew was reassembled at four A.M. on a Sunday, so there would be no curious audience of glass-workers to make the still-inexperienced ladlers nervous. They were alone except for a cleaning crew working in the steel girders and trusses of the roof, oblivious to the critical work on the floor beneath them. Dust and debris tumbled down into the open mold. Shouting and sign language finally routed the cleaning crew to another area of the building; a worker used a jet of compressed air to blow the debris out of the mold; and the ladlers started up again.

  This time the cores held in place, but a cascade of bubbles rose from the surfaces of the mold, spoiling the disk. McCauley asked the mold makers and the men who had filled the tank to make sure that all materials in the mold and the batch of glass were exactly the same as they had used in the successful practice pours. An expert on gas analysis from the Corning labs punctured one of the larger bubbles in the disk and collected a sample. His gas spectrometer reported that the gas was from the combustion of a petroleum product.

  For McCauley problem solving was a Sherlock Holmes game. He talked each member of the crew through the procedures they had used in the trial molds and the most recent efforts. Where had they gotten the bricks? Which cement had they used for the molds? Which lots had the material for the glass melt come from? The only change he found was that the early molds had been cleaned with a vacuum cleaner. The technicians couldn’t get the nozzle of the vacuum cleaner between the cores that had been glued in place for the ribbed disk, so they had used compressed air to blow the mold clean.

  Why would the switch from a vacuum cleaner to compressed air make a difference? The exhaust from a vacuum cleaner is expelled outside the machine. With compressed air a trace of oil from the compressor could have gotten into the airstream and ended up on the mold lining. When the molten glass hit the mold, at temperatures upward of 1000°C, a microscopic trace of oil would be enough to emit gases into the molten mixture, producing bubbles. For all future castings, McCauley ordered, the molds would be cleaned only with vacuum cleaners.

  By April the first annealing oven was ready. The trial disks, which had been slowly cooling under Sil-o-Cel insulation, were consigned to the oven and held at 475°C for ten days, then cooled one degree Celsius per hour until the oven could be opened. At the unveiling McCauley found an opalescent twenty-six-inch disk, the strains in the glass obvious from the cloudiness, and a broken thirty-inch ribbed disk. The residual stresses at the rim of the smaller disk were eighteen times what he had anticipated. McCauley recalculated his annealing constants and concluded that the temperature they had used was 45° too low. The break in the ribbed disk, he concluded, was from a tiny fracture produced when the mold cores were removed from the disk. Reheating in the oven had aggravated the fracture, and the annealing temperature had been too low to refuse the glass. Corning was learning about handling the glass, but after six months of experiments, they still had no disk to offer the Observatory Council.

  In May, McCauley tried again. This time he replaced Woods with two experienced glass ladlers from A Factory who had seen him struggling with the big ladles and thought they could do better. The ladlers cast three disks all at once, another solid disk, and two thirty-inch ribbed disks. One of the ribbed disks went straight into the annealing oven. Two weeks later it emerged, free of bubbles and with stresses within the expected range. The only flaw was that at one point the glass had not completely filled a portion of an outer circular rib. Knowing that the outer portion of the disk would be ground down to a slightly smaller size, McCauley sent it off to the Mount Wilson Optical Labs in Pasadena. The opticians found the material easy to grind and figure and pronounced it perfect in their tests.

  After the long months of frustration with GE, Hale, Anderson, and Mason were delighted by the progress Corning had made. In February, Mason wrote that conditions were “ideal” for getting Corning to do the work: Production was slack at their plants, they were interested in as much work as the Observatory Council would provide, and Day and the other Corning people had a “fine scientific attitude.”

  After some preliminary grinding on the thirty-inch disk, Hale summoned another meeting in New York City, where he gave Corning authorization to proceed to the sixty-inch disk, and if it was successful to go ahead on the one-hundred-twenty-and two-hundred-inch disks. No one asked for a formal contract. Three meetings and an exchange of letters was enough to agree on terms. Corning gave their “best estimate based on experience in the manufacture of disks less than 30” in diameter,” and agreed to bill monthly for “actual costs,” which meant their direct expenses plus overhead. If the disks were successful, they would expect an additional 10 percent profit. The confirming letter from Corning left an escape clause: “Since there is some uncertainty regarding success in making these large disks, either party shall have the right to withdraw at any time.”

  The thirty-inch disk from Corning, the first successful ribbed disk, took a good figure in the optical shop—the opticians’ way of saying that they were able to grind and polish it to a precise shape. More important, the ribbed back made the disk rigid, lightweight, and thin enough so that it did not suffer the slow adjustments to temperature changes of a solid disk. While there was no assurance that the Corning procedures would work on larger disks, with a successful ribbed disk in hand, the Observatory Council directed their efforts toward a ribbed two-hundred-inch disk for the telescope, with support mechanisms to maintain the disk’s shape.

  The ribbed back was essential. But what should the waffle look like? Hale had proposed the problem of achieving maximum rigidity for a given minimum weight of disk to the mathematical physicists Harry Bateman and Paul Epstein, and the engineers Von Karmann and Romeo Martel, all at Caltech. The four reported to Hale that the problem could not be solved. Arthur Day, surprised that Hale had given the problem only to men with no experience in the design of telescopes, wrote to Francis Pease and found that he already had a solution to the design problem for the mirror back—based on a series of wheel hubs, each of which would hold a support mechanism, with radial spokes between the hubs forming a rib structure.

  Pease was devoting much of his time to the telescope now. A dozen years before, when the one-hundred-inch was new and sensational, Pease and Anderson had conducted a research program on the new telescope, based on an idea of Albert Michelson’s, which used
a huge interferometer erected on the upper end of the telescope tube and prisms near the focus to measure the actual disks of distant stars. By 1930, with Hale’s encouragement, Pease had built a fifty-foot-long interferometer on Mount Wilson to provide even greater angular resolution. In 1932 he finished one last experiment, a refined measurement of the speed of light proposed by Michelson, and then put much of his own research interests aside to concentrate on the engineering of the two-hundred-inch telescope.

  There still had been no public announcement of the work going on at Corning, but Dwight Macdonald, at Fortune magazine, had his antennae out, picked up some signals, and did an article on the progress of the telescope and the involvement of big business in the project. Max Mason, running interference for the Observatory Council, tried to get Macdonald’s draft toned down. The phrase “which G.E. had done at cost,” he wrote Macdonald, “is rather dragged in by the hair. Everyone mentioned did work at cost or for nothing at all.” He also downplayed what Macdonald had called the “mystery” of the transfer from GE to Corning. The only mystery, he wrote, was the whole telescope, “because no one is willing to bet 100 to 1 that the 200-inch disk can be made.”

  With the cat out of the bag, Hale had no choice but to announce the work at Corning. The announcement was deliberately low key, focused on the switch from fused quartz to Pyrex rather than the change from GE to Corning. The press release caused little stir. By midsummer 1932 Americans had their own problems. Those who could afford it could live in a boarding house for a week for $3.50, could buy breakfast for $.10, or ride a trolleycar for $.05. Many couldn’t afford even those prices, and lived in tin Hoovervilles or flophouses, where as many as three or five hundred double-decker beds were crammed into spaces that reeked of humanity, and where long lines queued up for meals of soup and bread. Andrew Mellon might call the depression a “hiccup” in the nation’s economic life or announce that “America is going through a bad quarter of an hour,” but families who scrimped with their meager savings or relief funds that averaged as little as $8.00 a week, trying to keep out of the poor-houses and Hoovervilles; and those who were forced to choose among heat, food, or clothing, had enough on their minds not to worry about the squabbles among the companies building a “giant eye.”

  The temporary respite from the press was a relief in Pasadena. The signs of progress from Corning meant that the design program could shift back into high gear. For years everyone had pushed their pet ideas about the mirror. Many questions were settled, but enough remained for Hale to ask Mason to head the design group for the mirror. It was unusual to have the head of the funding organization sit on a working committee of the project, but Max Mason was an unusual foundation president. As a physicist, and in his wartime research on acoustic detection at the predecessor of the Navy Sound Labs in New London, Connecticut, Mason had been more comfortable with solid applied problems, whether the wartime adaptation of Broca tubes to submarine detection or the later engineering problems of the telescope, than with the administrative and diplomatic duties of a foundation.

  From Hale’s perspective Mason was a superb choice to pull the mirror design together. He was respected as a physicist, which meant the scientists would listen to him; he had experience with applied engineering in his work on acoustic devices, which meant the engineers would listen to him; and he was enough of an outsider, working in New York, to be beyond the inevitable disputes of territory and personal preferences that arose between opticians and astronomers, Mount Wilson staff and Caltech staff, engineers and observers, telescope designers and glassmakers, and even among the scientists, between the spectroscopists and those who photographed distant objects.

  The whole telescope depended on the mirror. When good news arrived from Corning, every other committee perked up. In 1929 the Site Committee had placed ten of the small telescopes Porter had designed at ten different locations in Southern California and Arizona. Each telescope was equipped with a high-power eyepiece that enabled a relatively untrained observer to measure the atmospheric oscillation of star images under a high magnification and to grade the images with the method Anderson had devised. The seeing tests were supplemented with weather records, using instruments borrowed from the U.S. weather bureau to record the extent of cloud cover and the number of days with sunshine. A year later the search was down to seven sites. The committee concluded that the summer rainy season in Arizona and the low winter temperatures there were unfavorable for a telescope site. The criterion that the site be within a few hours’ ride of Pasadena reduced the list to three: Horse Flats, fifteen miles north of Mount Wilson; Table Mountain, twenty-five miles from San Bernardino; and Palomar Mountain.

  By 1932 sunshine recorders and cameras were still in place at the other sites, but the preponderance of test figures were from Palomar. The initial readings for Palomar had been compiled by Mr. and Mrs. William Beech, who owned a ranch in the French Valley section with a cabin. Later a twelve-inch telescope was installed on the mountain, and observers from Caltech or Mount Wilson would go down to take more precise figures. Ferdinand Ellerman began a long series of observations to record the seeing through the seasons. He was troubled by the fog and noted that in optimum conditions the seeing at Mount Wilson was better than Palomar. But the average visibility—in terms of weather, light pollution, and atmospheric steadiness—was superior. Early in the search Hale had written that “results thus far point to Palomar as the most promising.” Even as bouts with his private demons pulled him further from the project, the prophecies of the master builder of telescopes remained persuasive.

  McCauley considered the thirty-inch disk only a partial success. The poured glass hadn’t filled every crevice of the mold, leaving part of the ribbed structure unmolded. Again McCauley set off on one of his Sherlock Holmes sessions, exploring every possible cause before he concluded that the glass in part of the mold had begun to set before the mold was completely full. These disks used far more glass than Corning was used to pouring. To avoid worse complications with the larger disks, he decided that the disks would have to be poured into a heated mold. Direct heat, through the insulating refractory brick of the mold, was impractical. McCauley’s answer was an “igloo,” a beehive-shaped dome over the mold, lined with nichrome heaters. The igloo would be preheated and maintained at temperature throughout the casting process to keep the mold and the poured glass hot enough to prevent blank spots. A doorway in the igloo would provide access for the ladles of glass.

  Working late at the dining room table, he transformed the ideas into sketches. The engineers and draftsmen turned the sketches to working drawings, and the masons, millwrights, and electricians built a new annealing oven, big enough for a sixty-inch disk, and an igloo for the mold, sandwiched into the space between two of the huge tanks used for ordinary Pyrex production work. The mirror project was substantial enough that Amory Houghton authorized a general policy of releasing workers from the various trades within the factory as needed. Corning reticence and Hale’s injunctions against publicity were enough to restrict general announcements of the new project. Some factory workers realized what was going on, but Corning is a self-contained town. Word did not spread to the outside world. “We still enjoyed the quiet, undisturbing atmosphere,” McCauley recalled, “perhaps not yet fully appreciated, of working together in the privacy of our own back yard without fear of being watched.”

  When the equipment, including an overhead traveling hoist to move the igloo cover and ultimately the disks, was in place, McCauley got his crew together to cast a sixty-inch disk. They rehearsed the whole procedure until the steps of opening the door in the heated igloo to receive the ladles of hot glass went smoothly. As ladles of glass were added, the level of glass rose in the mold, filling the crevices between the cores that formed the ribbed pattern of the back of the disk. The heated igloo kept the glass molten long enough to fill every corner of the mold. Another problem seemed solved.

  Except that it wasn’t that easy. The glass was barely over
the tops of the molds when a rib form popped loose and bobbed to the surface of the molten glass. Others followed, ruining the neat geometric pattern of the back of the disk. McCauley stopped the pour.

  A few days later the masons had built a new mold, to stricter specifications: stronger dowels for the cores, precise procedures for the cementing of the cores and the coating of the interior with silica flour. The masons were told to treat the entire mold as fragile when they vacuumed out debris. A try at casting the disk in the new mold went smoothly until the glass was almost at the required thickness. When the last ladle of molten glass was poured into the mold, a single core broke free and popped to the surface.

  McCauley reacted quickly. Instead of calling off the pour, he ordered the ladlers to finish and then to use tongs to fish out the errant core before the disk was consigned to the annealing oven. After the disk was annealed, the back could be ground to produce the precise geometric shape of the missing core form.

  When the surface of the molten glass leveled, the overhead crane picked up the annealing kiln. The crane order had been for a special “slow moving” model, but it still raced across the track, the kiln cover hanging at a rakish angle. A glassmaker jumped onto the kiln to balance it, then had to hang on as his weight sent the hoist rolling down its track, high over the floor. It took three husky millwrights finally to stop the swinging load and stabilize it. Plenty of hearts missed beats, but the Keystone Kops antics miraculously didn’t harm the disk.