The Perfect Machine

by Ronald Florence

  Besides the crating work, the summer seemed quiet in Corning. There were still back orders for disks, but with the two annealing ovens occupied with the two-hundred-inch disk and an 86-inch mirror blank for Heber Curtis at the University of Michigan, McCauley directed all efforts at smaller mirrors that didn’t need the big ovens. In addition to some specialized mirrors for solar telescopes, the publicity around the pouring of the big disk had generated an unexpected flurry of orders for one-tenth-size replicas of the two-hundred-inch mirror. Caltech wanted one for a mockup of the telescope to test designs. The other orders were from individuals and institutions that apparently reasoned that what was good enough for a two-hundred-inch mirror was good enough for a 20-inch one. In fact the ribbed back, necessary for the bigger mirror, served no purpose on a smaller one, except to present a headache to the Corning engineers: The tiny mold cores were too small to enclose metal mold anchors. McCauley concluded that they would need a “sculptor” to carve the mold out of solid blocks of insulating brick.

  The first week of July saw steady rain in Corning. The ground turned sodden, and the normally tame Monkey Run filled to its banks. Some puddling showed up here and there in the caves under A Factory, but it was no more than the usual sporadic flooding from spring runoff. McCauley was glad he had taken the precaution of building the annealing oven, where the precious two-hundred-inch disk was halfway through its annealing schedule, well above the highest high-water mark. Still, his family noticed his nervousness as the rains continued. It was only at the end of the week, when the rain tapered off, that he relaxed again.

  Then, on Saturday, the rains returned to south-central New York State. The downpour was steady and heavy, from early afternoon through the night. On Sunday morning McCauley’s son Jim woke up in his cabin in the Finger Lakes to cries of “Help!” from a neighbor whose boathouse had floated away. The lake had risen six feet overnight. In Corning the Monkey Run became a raging torrent. With the ground of the hills above the glassworks saturated, the runoff coursed down to the river below, overflowing the banks of the Monkey Run. The Chemung overflowed the dikes along the bank in front of the factory and covered the roadbeds of all three bridges over the river. It was the worst flooding in forty years.

  By Monday morning, when the early shift showed up for work, the factory yard on Walnut Street was a shallow lake. To get into the factory McCauley had to skip from one “island” to another. He ran down to the caves, where the controllers and thermostats for the annealing ovens were installed on their raised platform.

  Water was rising on the floor of the caves, coming in from the overflow of the Monkey Run to the south. On the north end of the factory, closer to the river, the water was rising even faster. A quick inspection revealed that the river had risen over the drainage ports in the concrete dikes; the holes that normally drained the caves into the river were instead bringing the river into the caves. By midmorning the entire floor was covered with water, and the level was rising rapidly.

  McCauley got the factory foreman to issue an all-hands call for men to build a restraining wall of sandbags, brick, and concrete to isolate the section of floor below the controllers and transformers for the annealer. With dozens of men working, the wall went up quickly. But as the double line of sandbags cut off the flow across the floor, the hydraulic pressure of the rising water in the rest of the factory forced jets of water up through cracks in the concrete floor under the annealer and controller.

  An emergency call to the Corning Fire Department brought men and pumpers. When someone on the phone mentioned that the effort was needed to save the two-hundred-inch disk, the fire department sent every pumper it had, filling the passageway outside the west side of the factory with red fire engines and a maze of black hoses. Alas, there was no place to pump the water except back into the river. Even with every pumper the fire department could supply, they couldn’t get ahead of the steadily rising water. Down on the floor, electricians, masons, laborers, and scientists worked side by side, standing in water up to their knees as they wrestled with sandbags, bricks, and mortar. McCauley remembered the Pathé news cameramen a half year before, who had been so anxious for action shots.

  All Corning’s men and equipment were no match for the floodwater. As the water level marched steadily toward the high-water mark left from the Great Flood of 1918, McCauley knew the battle was lost. The only hope was to cut off the electrical power to the transformers before they shorted out and somehow move them to dry ground and reconnect them in time to maintain the annealing schedule on the glass disk. The disk was down to 370°C. If the power were cut off, the insulation of the annealing oven would limit the rate of cooling for a short while. How long they could go without power without introducing strains into the glass—twelve hours, twenty-four, forty-eight—was anyone’s guess.

  The laborers who had been building walls of sandbags were put on air hammers on the floor above the transformers. Other workmen cleared away the production equipment for molding nursing bottles in the factory area above the transformers while the hammers pounded at the heavy reinforced concrete floor. The call went out for more jackhammers, but rising water in the compressor room cut down the air supply, limiting the number of hammers they could keep in operation. The water in the cave began creeping up the sides of the transformer cases. If the water reached the lids, it would flood the interior of the transformers, displacing the oil and rendering them useless. Replacement transformers would take weeks, maybe months, to order. During that time the disk would cool uncontrollably.

  McCauley had the laborers concentrate all the jackhammers on a single hole. A crane was rigged overhead, ready to lift the transformers to safety. In the cave below, electricians on the scaffolding disconnected the reactor units that controlled the flow of current to the oven. They got sixteen reactors out before the water was too high to work. Just over their heads, as many men as room permitted hammered at the concrete floor. As soon as a man tired on a jackhammer, another took his place. No one had to be reminded of the consequences of failure. Everyone there had seen the world rush to Corning to witness the casting of the great glass disk. This was the piece of glass that had made Corning famous.

  In the midst of the chaos, McCauley did some rapid calculations and concluded that one transformer could supply enough current to complete the annealing schedule; there would be no margin for error and no backups, but it would just work.

  The water was lapping the lower edge of the transformer lids when the hole in the floor was finally large enough to lift the first transformer to safety. The bottom edge of the lid was already wet. A few more minutes and the water would cover the top, saturating the transformer. A cheer went up as the crane hoisted the first transformer clear. Through the hole in the floor, men could see oil seeping out of the other two transformers as the water lapped over the lids and flooded the windings inside.

  When the first transformer—the only one they had rescued in time—was lowered to the floor, McCauley saw that the oil gauge on the side had been broken off. A sample of oil drawn from the hole was contaminated with water. The other two transformers were already under water.

  It was morning outside. McCauley and many of the workers had worked all night.

  All day Tuesday reporters called Corning, eager for stories about the fate of the big disk. McCauley, too busy to take calls, told the publicity office that there had been no damage to the disk, that in fact the water had never been within five feet of the disk. That was all he said, and all they printed.

  On the factory floor, laborers suspended the waterlogged reactors and controls outside a heated glass tank to dry. Even the one salvaged transformer couldn’t be used as it was. McCauley had a pump hooked up to pump the oil out of the bottom of the transformer case through a centrifugal separator that would separate off the water; the oil was then pumped back into the top of the transformer.

  McCauley finally went home Tuesday evening. He had been working without a break for thirty-six hours. T
oo anxious about the disk to sleep, he worked with his little drafting board at the familiar round table. He guessed that it would take at least another twenty-four hours before the reactors, controls, and transformer were dry enough to use. It might be as much as another thirty-six. In seventy-two hours without power, the temperature of the disk would drop seventy degrees, to 300°C. Would the glass survive that interruption to the annealing schedule?

  All day Wednesday, McCauley went back and forth from his calculations to the cleanup and salvage work. No one in his family had ever seen him so jittery. It took forty-eight hours to get the two flooded transformers out of their position. By then the water in the caves was receding, draining the transformers. Plant electricians estimated that they could be dried with current applied to the primaries, but it would be a long, slow operation.

  The question of what to do with the annealing schedule was McCauley’s to answer. There were no precedents for restoring an annealing schedule. At best it was a black art, as much guesswork as science. The only thing clear was the consequences. Too much heat or too little, heating too fast or too slow, could leave strains in the glass, destroying the usefulness of the disk. McCauley didn’t say much about his real worry: Could the disk survive the sudden drop in temperature intact? Or would they open the annealing oven, in November, to find the glass cracked?

  Thursday, when the transformers and reactors were dry, and power was restored, McCauley raised the temperature of the disk to 370°C, where it had been before the power was interrupted. He had decided to hold it there for five days, then resume the original schedule. His new schedule was a guess. All McCauley knew for certain was that they had lost eight days from the annealing schedule, and that after six months of controlled cooling, dropping by a carefully measured 0.8°C per day, the temperature of the disk had dropped seventy degrees in three days.

  For the next five months, every day when he checked the thermocouples on the oven, he thought about the flood and wondered about the disk hidden inside the annealing kiln.

  In October the eighty-six-inch disk for Heber Curtis at the University of Michigan was cool enough to come out of the smaller annealing oven. The oven had gone through the same power outage and sudden cooling as the bigger oven with the two-hundred-inch disk. The disk emerged intact—it had not broken during the power outage—but it was so filled with glass faults that McCauley classified it R.R.R. (reject requiring replacement). Curtis wasn’t disturbed by the news. He suggested that they make the replacement a ninety-eight-inch disk.

  Curtis, who would get a bigger disk, might be sanguine. McCauley wasn’t. The eighty-six-inch mirror was made from the same batch of glass, in the same 3A tank, that had been used to cast the two-hundred-inch disk. The glass had looked fine when they poured the first two-hundred-inch disk, but in August, when the tank was cooled off after the last of the big mirrors had been poured, the cullet that was broken out of the tank contained streaks of devitrification. The streaks seemed to mark the outlines of the ladle skins—suggesting that the glass that had been returned to the tank during the ladling, to keep the glass level up so the subsequent ladling efforts could reach the molten glass, had been contaminated with some impurity. What if the impurities also affected the two-hundred-inch disk?

  By mid-October, the natural cooling rate of the insulated annealing oven finally matched the rate that had been maintained by the electric controllers. The disk was cooling itself without help. The day-and-night guardians of the controls were moved to other jobs, and the power was turned off. On October 17, McCauley lowered the screw hoist carrying the disk one and a half inches, increasing the rate of cooling. The day of unveiling was rapidly approaching.

  After the disasters of the early publicity, McCauley knew he had to explore the unknown alone. On October 25, a Friday, he waited until all office and laboratory personnel in A Factory had gone home for their dinners, then went down into the factory cave alone. With the heaters in the annealing oven turned off, the cave area was dark, so he carried a flashlight. At the disk he stood motionless, listening to make sure he was alone in the cave, before he pushed the hoist’s down button. When the disk had lowered just enough for his slim 150 pounds to fit, he stopped the hoist and crawled up onto the still uncomfortably warm glass, keeping his thumb off the flashlight button until he was entirely inside the oven.

  When he switched on the beam, he saw a strip of clear, solid glass. The disk hadn’t broken.

  Then, as he moved the beam of the flashlight, the light came back in scattered reflections, as if from broken glass. “No!” he thought. “Not this one, too!”

  McCauley crawled toward the reflections. The surface of the disk looked as if some Paul Bunyan had embedded a huge crowbar several inches into the glass while it was still soft, then waited ten months to pry it loose. His mind raced, wondering what could cause scars like that. Could it have been the power failure during the floods? Was the disk ruined?

  He scanned the disk with his flashlight, looking for more wounds, then crawled over the surface toward each broken reflection, supporting himself on his knees and left hand, while he traced over the glass with the flashlight in his right. He found a second, a third, a fourth—nine in total. He sat down at the ninth scar, exhausted. Years of planning, two pours, almost a year of annealing, and it came to this. He felt as if his life had come to an end with nothing accomplished.

  As he sat on the disk, his eyes roved upward. The cover of the kiln had been fabricated to hold a layer of insulating brick over the surface of the disk, matching the insulating qualities of the mold underneath. Above the scar in the surface of the glass, he could make out a welded junction of beams in the cover. Could this have been Paul Bunyan’s crowbar?

  The physicist began calculating. When the molten glass disk was raised into the oven, the heat of the glass would make the side of the cover closest to it expand. The layer of insulating brick would shield the upper layer of the cover from the heat, and the uneven expansion would make the cover bend downward toward the glass, perhaps far enough for the steel beams to touch the surface. The explanation almost fit. But there were only three welded areas on the lid. How would they account for nine scars on the disk?

  The physics was easy. Common sense took longer. McCauley finally realized that in his anxious search for scars he had crawled around the disk three times. He had “found” the scars on each lap, magnifying the problem by three. Small consolation; even three scars, in what was supposed to be a perfect disk, were three too many.

  As he maneuvered around to climb out of the kiln, he saw the head and shoulders of Ralph Newman peer under the oven cover. Newman, who had worked on the mirrors from the beginning and had every right to be in the kiln area, was surprised at the vehemence of McCauley’s order to maintain “absolute silence” about what he had seen.

  With the kiln closed, that night McCauley thought through the options. It was clear that the Observatory Council should not be expected to pay for another casting. The telescope project was already behind budget and schedule, and as much as the Corning Glass Works had reveled in the attention of the world when they cast the disk, McCauley wasn’t eager to preside over the announcement of a Corning Glass Works failure. A second option was to reheat the disk to level its surface, then reanneal it. But the experience with the first disk, which had been reheated to smooth the rough surface left by the floating cores, hadn’t been encouraging. The reheating had introduced fractures to the ribbed supports. And reannealing meant another year in the oven, with the risks that entailed.

  There was one other option. The disk had deliberately been cast deeper than required. What if the surface were ground down far enough to remove the surface scars? McCauley calculated that the resulting disk would be thinner than the original plans had specified. Would it be too thin? The only evidence on the rigidity of mirrors with ribbed backs was based on much-smaller mirrors. The test results from Pasadena had all been good. The mirrors held their shape better than the designs had
anticipated. The two-hundred-inch mirror would have active mirror supports in the pockets in the back, helping it hold its shape against gravity. If the mirror supports worked, the thinner disk might even be better, because it would react faster to temperature changes. As he calculated and considered the alternatives, McCauley was convinced that grinding the mirror down was the only choice.

  McCauley wasn’t the only one thinking about the mirror. After the orgy of publicity at the pouring of the first disk, the telescope, and astronomy, had become a regular topic for newspapers, radio broadcasters, and the weekly and monthly magazines that supplied news and entertainment to Americans. Reporters, prompted by their tickler files, called constantly for progress reports. “In December, you said the annealing would be finished in October. How soon can we see the disc?” Every potential disaster, like the flood, brought another round of calls from reporters who hoped for a fresh angle on the mirror.

  In the depths of the depression, the telescope had become a staple of the news, a symbol of scientific and technological progress. When Edwin Hubble’s search for larger red shifts led to the idea of an expanding universe, journalists seized on both the connections to the famed Albert Einstein and the loose ends in Hubble’s evidence. Few reports on Hubble failed to assure their readers that the great two-hundred-inch telescope, still under construction, would solve the dilemmas. The humorists, too, had made the telescope project grist for their columns in magazines like The American. After the “minor problems” with the first casting were announced, Robert Benchley sarcastically explained to his readers that what had gone wrong was that “40,000 pounds of glass is too much glass.” The telescope disk became a theme for Benchley, who picked up on each announcement of another mirror being cast at Corning: “If you ask me, they have got started making gigantic glass lenses up at Corning and can’t stop. And California is being made the sucker.”