The Perfect Machine

by Ronald Florence

  Hale knew that his withdrawal from the project left a vacuum. John Anderson would increasingly be concerned with the figuring of the mirror. Sandy McDowell was efficient at arranging and supervising contracts and pushing along the work on the mountaintop, but he had no experience in astronomy or with telescopes, and his aggressive management style was increasingly resented by the scientists on the project. Walter Adams had his hands full with Mount Wilson, and Millikan, though he had never taken the title of president of Caltech, was running the institution, his time taken with fund-raising, faculty raiding, and aggressive promotion of the school. There didn’t seem to be anyone who could step into George Hale’s shoes.

  To complicate the question, in November 1935 Max Mason announced his retirement from the presidency of the Rockefeller Foundation. Mason had taken over at the Rockefeller Foundation just after the grant for the telescope had been announced. He had presided over the reorganization of the foundation and boards, consolidating all science activity in the Rockefeller Foundation. The payments to Caltech for the telescope came first from the IEB, which had made the grant, and then from the GEB, which had taken over responsibility for the grant after the IEB funds were depleted. But even as the checks came out of those accounts, it had been Mason and the men he hired at the Rockefeller Foundation, like Warren Weaver, who supervised the grant. Mason had fended off the interference of Merriam at the Carnegie Institution, handled some of the trickier negotiations with GE, recommended McDowell to supervise the construction work on the telescope, and served as a sounding board for Hale.

  In 1932 Hale had even asked Mason to be chairman of the committee on design for the mirror of the telescope. It had been a largely symbolic position, but Hale felt that Mason—as a theoretical and experimental physicist—could pull the work of Pease, Day, McCauley, and others together. It had been an unusual step to ask the president of the foundation supporting the project to participate actively in the project. Usually grant recipients do all they can to keep the funding source at arm’s length, lest the leverage of funding influence a project. But Mason’s counsel had been so valuable, and his authority as a scientist and administrator of science and head of the Rockefeller Foundation so effective, that Hale treated him as a member of the Observatory Council in all but name.

  Mason’s retirement gave Hale an idea. In March 1936 Hale suggested that Mason come to Pasadena and accept dual appointments as a research associate at Caltech and vice chairman of the Observatory Council. Confidentially it had been agreed that with George Hale’s health rapidly failing, Mason could serve as the chairman of the council in case of Hale’s absence or incapacitation. Mason agreed, and by October 1936 he was settled in Pasadena, and already getting McDowell’s memos on pay scales, organization charts with side boxes and arrows, copies of memorandums on contract details, and showing copies of memos on myriad other details. Mason still consulted with Hale whenever Hale was able to receive visitors, but before long he wrote to Ray Fosdick at the GEB that the work was going well “without Hale.”

  Mason had been chairman of the Observatory Council for just two months when another crisis threatened the entire project.

  24

  Crisis

  George McCauley had been looking forward to the calm of a Corning autumn. There were still a few back orders of astronomical disks to fill, but the crews were experienced, the procedures practiced, and the technical details established. Although they were still made under the auspices of the research department, telescope disks had become a production process at Corning. McCauley, who had developed the processes and refined the technology, looked forward to answering the mountains of mail he had received, writing up articles for the journals on the engineering and physics issues involved in making the disks, and responding to the many requests for lectures on the disks, especially the now-famous 200-inch disk. His expectations were short-lived.

  At the optics shop on California Street, Brownie and the crew were steadily grinding down the surface of the big disk. Anderson and Brownie had been buoyed by McCauley’s calm assurances during his visit in early July and expected that at any point the grinding would finally work through the last of the troublesome checks and fractures. Some weeks the work went well. When Brownie and his crew washed the disk down on Friday afternoon, in preparation for Anderson’s careful examination, there were times when many of the troublesome fractures of the week before had been washed away in the slurry of carborundum, glass, and water.

  But even in the best weeks, Anderson would find new fractures, deeper in the glass, especially in a dark stria that showed up halfway between two of the pour points. He began sending McCauley regular reports, often with photographs of the fractures they had discovered.

  The photographs wrecked McCauley’s plans for a quiet fall. He had believed his own assurances to Anderson early in the summer. There hadn’t been comparable troubles with the other disks, including the 120-inch disk. What had been so different with the two-hundred-inch disk?

  The continued reports and photographs got him worrying. Maybe this disk was different. His first guess was that either the shutoff of heat to the annealing oven during the flood or the contact of the cover plate with the surface of the disk was to blame. Yet his tests with a polariscope after the annealing had showed minimal strains in the disk; as frightening as the shutoff of power during the flood had been, it didn’t seem to affect the annealing. And as it became clear that the fractures were distributed deeper in the disk than the surface wounds from the contact with the cover of the annealing oven, McCauley knew he needed a different explanation.

  He recruited Ralph Newman, his lab coworker from the days of the first experiments for the mirror project, to work with him. Painstakingly they reconstructed the pouring procedure. The proximity of many of the fractures to the three pour points was suggestive, but the nagging fact that at least three big disks produced by the same pouring process, including the 120-inch mirror, had been successfully ground with no checks or fractures, wrought havoc with every explanation McCauley tried.

  By December, Brownie and his crew had used close to five tons of carborundum in the rough surface grinding. More than two tons of glass had been ground off the disk. The surface level at the edges was approaching the final planned thickness of the disk. On Friday, December 11, Brown gave the disk a thorough cleaning. The next morning, Anderson and Max Mason came to the shop to inspect the disk with Brown. There were fewer fractures in the glass than a month before, but there was at least one relatively large check that was not near the flow points. In a Monday letter to “Dear Mac” (McCauley), Mason suggested that it would be useful to simulate the flow procedure of the pour with a model to determine the cause of the fractures.

  It was an interesting suggestion. In Mason’s own research at the navy underwater research labs in New London, the modeling of flow patterns was an essential test procedure. But water flowing past the hull of a submarine or a surface ship is easier to model than molten glass. Glucose and water, the closest material McCauley knew of to simulate the flow of molten glass, was not temperature-sensitive enough to mimic the behavior of Pyrex.

  McCauley tried to appear calm and reassured Mason and Anderson in California that the checks would eventually disappear before they ground the glass down to the final curved surface of the mirror. This time McCauley’s assurances weren’t convincing. In January, Mason wrote Houghton—the mail was marked PERSONAL and CONFIDENTIAL—to ask the cost and delivery date for another two-hundred-inch mirror. McCauley agreed to go out to California for another look. All discussion of the state of the disk, in Pasadena and in Corning, was kept secret. After the incessant attention of the press during the unveiling of the disk and the journey across the country, both Pasadena and Corning were wary that a reporter would seize on a mention of a fracture in the disk for banner headlines: GREAT EYEGLASS FLAWED or THE EYE THAT COULDN’T SEE. And Robert Benchley was always waiting to let loose another barb:

  I hate to keep harping
on this subject, but what do they do with gigantic telescope discs out there in California—eat them? … 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. I go to California every year, and I have never seen any more glass that you could put a highball into. They must just throw the glass away when it finally, after weeks of publicity, gets out there.

  McCauley took another secret trip to Pasadena at the end of January, 1937. He and Anderson pored over the disk, examining the troublesome areas with magnifying glasses. When McCauley wanted to take samples of the disk back to Corning for analysis in their laboratories, Anderson said that it would probably be quicker if they did whatever tests McCauley wanted at the Mount Wilson laboratories on Santa Barbara Street in Pasadena. He phoned the lab, and a few minutes later, he and McCauley showed up with two samples bored from the disk, one of clear glass, the other from a cord that had developed checks.

  At Santa Barbara Street, while he waited for the spectroscopists to test the samples on their large-grating spectrograph, McCauley got into a conversation with the opticians who were working on various smaller mirrors. One older optician said that he had run up against checks like the ones McCauley described. On an optics bench, he showed McCauley the face of a mirror that was fully figured and polished, awaiting silvering. After the optician showed him where to look, McCauley could see a fine fracture, approximately an inch long, in the surface of the mirror. The optician assured McCauley that the fracture would present no problems in silvering the mirror and would not impair its performance. McCauley asked for the records on the disk, and found that it was Pyrex and had been made before the switch over to the 715-CF glass that McCauley had used for the bigger disks. It was scant consolation that the two-hundred-inch wasn’t the only disk from Corning that had problems.

  Before he left McCauley did his best to reassure Anderson that the fractures would disappear. He spent the whole flight back thinking about the results of the spectrographic tests. The two samples of glass were strikingly different. The spectra from the cord of glass with checks contained a higher content of alumina and soda than the sample from the clear glass—a clear indication of contaminated glass. The only possible source of that contamination had to be the refractory lining of the tank. But why did it affect only that disk? By the time he reached Corning, he was at “a new low in spirits.” Would the checks continue so deep into the blank that the two-hundred-inch disk couldn’t be figured? Would the whole telescope project have to be abandoned for lack of a mirror? The dread of affirmative answers to those questions haunted him.

  The only way McCauley could dispel his depression was by work. On the plane, and when he got back, he visualized the processes at work as a newly lined glass tank was filled with glass and fined. The procedure was cautious and precise; he was sure that the contamination wasn’t introduced then. That narrowed the possibilities to two points in the procedure: the agitation of the glass mixture against the walls of the tank by the ladles when they were filled, which might have stirred off contamination from the walls of the tank; or drainage down the walls of the tank as the glass level dropped.

  Every large disk had been poured with ladles, and the same men had handled the ladles each time. That seemed to rule out the agitation of the glass as the culprit. He went over the list of disks they had produced each year. The big disks cast in 1933, including the 120-inch disk, had all been successfully ground without showing checks. Two 60-inch disks for Harlow Shapley at Harvard had also been successfully ground.

  He narrowed his search to the disks cast after those successful ones. The first two-hundred-inch disk and an 86-inch disk for Heber Curtis had both been rejected, for reasons unrelated to the tiny fractures. Both blanks were still at Corning. When he got home, he stayed up late at the round table in the dining room, going over the records of the pours, making sure of his analysis.

  In the morning he examined the first two-hundred-inch disk, still resting on its bed of timbers in the steel crating room on the riverbank. Knowing what to look for, he quickly found the telltale fractures in the glass. The 86-inch disk that had been cast for Heber Curtis at the University of Michigan, which had emerged from annealing cracked, showed the same fractures.

  McCauley had been working on the mirror project for almost ten years. Now, after surviving the floating cores, the first false starts with the wrong glass, the grand public failure with the first mirror casting, a flood, an earthquake, the threats of preachers who had damned the project, and the trials of a journey across the country by rail—it seemed that they were using the wrong refractory material in the tanks, the wrong glass, or the wrong procedure.

  It had seemed so simple to copy the ladling procedures of the old window-glass industry. But window glass was ladled at a low temperature, and the tanks were so huge that the glass level scarcely dropped as the ladles of glass were removed. By returning the ladle skins to the tank, he had minimized the drop in level, even for the two-hundred-inch disk, to 10 inches. It was impossible to melt a tank large enough that the amount ladled out for a big telescope disk would not appreciably lower the level of the tank. When the glass drained down the walls there was a risk of contamination from the refractory brick. The older the brick, the more glass that had been recycled through the tank from ladle skins, the greater the risk. What had seemed a terrific production procedure—reviving the old technique of hand ladling—had backfired.

  After he received a letter from Max Mason reporting that although they hoped “the glass would behave better as deeper depths were reached” the fractures seemed to be “distributed throughout,” Houghton asked McCauley to prepare quotes of a price and a delivery date for a new mirror.

  The situation didn’t look promising.

  When the grant was first awarded, $6 million—the budget figure George Hale had pulled out of the air of his room at the University Club—seemed generous enough for every eventuality. Eight years later, after close to a million dollars had been expended on the GE experiments, and hundreds of thousands more for the series of mirrors from Corning, the machine and optics shops and the astrophysics building in Pasadena, and the site work on Palomar, the grant no longer seemed bottomless. Westinghouse had estimated the cost of the fabrication for the mount and tube at $0.37 per pound, a low, depression-era bid. For one million pounds of telescope mount, it would still come to $370,000, which did not include the cost of transporting the huge pieces to Palomar or assembling them there. The control system for the telescope, and the various eyepieces, spectrographs, electronic sensors, and other auxiliary equipment would also make demands on the budget.

  And even as the cost of the telescope mounted, there were new demands for additional instrumentation. The Schmidt camera on Palomar was performing beautifully on Zwicky’s searches for supernovas. The astronomers began to discuss the usefulness of a much larger Schmidt camera—as large as opticians could fabricate—to serve as a wide-area survey camera for the two-hundred-inch telescope. Hubble gave the proposed camera his blessing. When he had first searched for novas and Cepheid variable stars with the one-hundred-inch telescope, he had to guess at promising areas and photograph them with the narrow field of view of the big telescope. As he and Milton Humason expanded their search for distant galaxies, they were still using the same technique of guessing at the most productive areas to study. The eighteen-inch Schmidt camera had proved that the concept of a widefield camera worked, but the limiting magnitude and image size of the small Schmidt camera weren’t large enough to serve as a survey camera for the two-hundred-inch telescope. Hubble urged that they build as large a Schmidt camera as possible, preferably a fifty-inch f/2 instrument.

  The astronomers and the Observatory Council all enthusiastically endorsed the proposal, and Max Mason sent feelers to his former colleagues at the Rockefeller Foundation to see if they would approve the expenditure for a large Schmidt camera within the terms of the grant, which allowed for “the pu
rchase of a site, and the construction of an observatory, including a 200-inch reflecting telescope with accessories, and any and all other expenses incurred in making the observatory available for use.” Mason was sure he could get the new Schmidt telescope approved, but money spent for a big Schmidt camera came out of the same $6 million budget. Would there be enough left to pour, anneal, transport, and do the preliminary grinding on a new disk for the two-hundred-inch telescope—which might prove no better than the disk they had? How much was a reserve mirror worth in the now-tight budget?

  George Hale had been able to discuss questions like that one with Mason. Now Mason had crossed over to the Observatory Council, to being the grantee instead of the grantor. His relationship with his former colleagues at the Rockefeller Foundation was more personal than consultative. Their correspondence, sometimes marked PRIVATE, interspersed comments on mutual friends or fine wines and cognacs and armagnacs into the commentary on projects. When Mason reported on developments on the telescope project, he mentioned that they were still finding fractures at the point when all traces of fractures and checks should have been long gone. But the mention was quickly glissaded over in favor of a discussion of the difficulty of getting a decent 1926 Chambertin.

  With no one to ask for advice, and with the pressures of a dozen other aspects of the project pressing on him from all sides, Mason set the figure they would be willing to pay for another disk at fifty thousand dollars, exactly half of McCauley’s estimate of Corning’s cost, not including overhead or profit. If Corning could quietly produce a new disk for fifty thousand dollars, the council would go ahead with the order. In private correspondence Mason negotiated with McCauley and Houghton, pointing out that the problems with the disk weren’t due to the shutoff of power during the flood or the earthquake—the sort of “Acts of God” that are exempted in warranties and guarantees—but were instead striae in the glass from contamination with alumina from the walls of the melting tank.