The Perfect Machine

by Ronald Florence

  This time Ellis was confident. The latest disks to emerge from his furnace, though flawed, had been a great improvement over the earlier efforts, and the efficiency of the process had improved dramatically. At the beginning of 1929 the spray apparatus was depositing one cubic inch of quartz per hour while consuming two hundred cubic feet of gas. By the end of the year the nozzles were laying down two hundred cubic inches of quartz per hour and consuming only six hundred cubic feet of gas. A mirror could be built in one to 1.5 percent of the time the process originally required, which meant that the time during which the furnace and disk had to be maintained at the torturous temperatures was substantially reduced. Ellis had finally abandoned hydrogen as a fuel, because it would have required a gas plant larger than they were willing to build. Dissociated ammonia wasn’t hot enough, so they had begun experiments on alternative fuels, based on research done at the Department of Agriculture, Du Pont Ammonia, Nitrogen Engineering in New York, the Fixed Nitrogen Laboratories in Washington, and their own experience in Schenectady and Lynn. The most promising fuel was cracked butane, which could be obtained in carload lots, evaporated at any desired rate, and promised a hotter flame than any fuel they had yet tried.

  As Ellis and Thomson saw it, the entire development was progressing predictably. Ellis was “delighted” when the opticians in Pasadena found 80 percent of the surface of the last mirror shipped satisfactory. Ellis and Thomson considered the work that had gone before to have been experimentation, trials with different materials and procedures. The difficulties they had encountered were preproduction problems that could be expected in any experimental process. Although they hadn’t yet produced a fully satisfactory mirror, that was almost to be expected, and wasn’t really a problem in any case, since none of the smaller blanks were meant for use in the telescope. GE, and particularly Elihu Thomson, had been famed for turning experimental ideas into working products. The company’s claim to fame, and its prosperity, had stemmed from the translation of the inventions and discoveries of Edison and Thomson into routine production processes. Ellis was following the company tradition.

  What Ellis and Thomson never quite understood was that Hale, Anderson, the Observatory Committee in Pasadena, and the astronomers who were waiting for the telescope were not after a perfected production process. All they wanted was mirror blanks—three sixty-inch blanks for the auxiliary mirrors, a two-hundred-inch blank for the primary mirror, and a one-hundred-inch blank they could grind into a flat for testing the primary mirror. They needed only five disks, good enough to be figured into mirrors for a telescope. So far, after two years of experimentation, 80 percent usable was the best GE had achieved on any disk. The opticians couldn’t make a telescope mirror out of an 80 percent usable disk.

  Without the mirrors, or at least some assurance that a satisfactory mirror blank could be fabricated, Pease, Porter, and Anderson could not move ahead on a design for the telescope. And without a design, the rest of the work on the project—the search for a site, the mechanical and civil engineering for the observatory itself, the preliminary work on instrumentation and auxiliary lenses, the electrical engineering and calculations for a control system for the telescope—was on hold. No one objected if the work GE was doing on the fused-quartz mirrors ultimately led to techniques for regular production of astronomical mirrors, if only they could produce the mirrors for this telescope.

  The halting progress in West Lynn never stopped GE from issuing a steady stream of press releases and articles for the trade and popular press. While Anderson and Hale were having trouble coming up with excuses for the lack of progress on a mirror, John W. Hammond of the publicity department at GE was sending out articles with titles like “Greatest Venture in Mirror Making Ever Attempted,” “A Great Magnifying Glass to Help Read the Story of the Stars,” and “Building a Looking-Glass to Mirror Unknown Stars” to any magazine that would print them. Thomson, although he had left the day-to-day running of the project to Ellis, never turned down opportunities to speak on the subject of the mirror and the telescope. One of his talks on the two-hundred-inch telescope, in December 1929, was broadcast on radio from Philadelphia.

  GE’s appetite for publicity offended the reticence of the astronomers in Pasadena, especially George Hale. Much that the GE publicity department included in their articles was wrong. They wrote that the smaller mirrors were needed as “finders” by which the heavenly bodies are first located; in fact the auxiliary mirrors were needed to focus the image of the primary mirror at different positions on the telescope, so the same telescope could be used for both deep-space research and for detailed spectrographic study of nearby stars. The GE publicists had no qualms about announcing that the new telescope would open an area of unexplored space thirty times greater than at present known, or that the most remote of the charted stars were 150 million light-years from the earth—both untrue statements. It wasn’t only the inaccuracy of the reports that bothered the astronomers. They were afraid that the newspapers and radio stations would read between the lines of the GE reports and speculate about the lack of progress on the project, and that the speculation would in turn feed the doubters like Harlow Shapley, H. L. Mencken, and others who enjoyed taking potshots at the California astronomers.

  As the effects of the depression spread, with daily reports of bank failures, soaring unemployment, breadlines, and soup kitchens, Hale and the others were also embarrassed and worried that the scale of spending on the telescope project would prove difficult to justify. Six million dollars was still a lot of money, and astronomers were already the butt of jokes and cartoons about stargazing and heads in the clouds. Men who had lost their jobs and were now selling apples and pencils on street corners to feed their families might be less enthusiastic about a $6 million telescope than the ebullient newspaper readers and radio audiences of 1928.

  The differences between the astronomers in Pasadena and Ellis, Thomson, and Swope at GE remained largely unwritten and unspoken. Theoretically, fused quartz would make the best possible mirror.

  If GE could produce fused-quartz mirrors that would eliminate the problems that still troubled the one-hundred-inch telescope, the differences in style between the huge eastern corporation and its publicity department and the tiny West Coast university, the embarrassments over premature or inappropriate publicity—even the cost overruns wouldn’t matter.

  On the early experimental disks, even on the twenty-two-inch disks, Ellis could in a pinch get together enough usable quartz by putting a couple of men on mortars and pestles, and grading the material with hand sieves. For the telescope mirrors, Ellis ordered quartz by the carload, had it ground in a ball mill in West Lynn, then used graduated sieves to separate the powder by particle size. Extremely fine particles, which would foul up the jets of the spray equipment, had to be filtered out by an additional process. Traditional separation methods, like air-blast filtration, didn’t work because the fine particles would become electro statically charged and cling to larger ones. Ellis finally instituted a wet-filter process, in which the fine material would be suspended in a solution that was then drained away. The procedure was excruciatingly slow. It wasn’t until late summer of 1930 that he was finally ready to begin spraying the first sixty-inch disk. The graded quartz waited in hundreds of large glass jars for the day when the spraying would start.

  Two full years had elapsed since GE had agreed to fabricate the mirror blanks. The original budget of $252,000 had long been spent. Hale and Anderson repeatedly asked for a new budget. “It is of course impossible to make an estimate that will be hard and fast as final cost of producing the mirror, including the 200 inch,” Ellis answered. His new ballpark figure was an additional $50,000 for the furnace and accessories, and $12,000 per month for materials and development work, for a period of eighteen months. The total of $266,000, added to the $308,000 they had already expended, made a grand total of $574,000 for three sixty-inch mirror blanks, a one-hundred-inch mirror that could be ground to a flat for testing,
and the two-hundred-inch mirror. The price tag for the mirror blanks had doubled in two years.

  If the actual progress on the blanks was slow, Ellis and Thomson were still ever ready with ideas. Ellis drew up his own ideas for a mounting for the telescope—a project on which Pease and Porter had been working for years—and for the kind of facility in which the disk should be ground, polished, and tested. To minimize vibrations and temperature variations, he wanted the room entirely underground. He had suggestions on how the two-hundred-inch disk should be moved to a optical shop, equipment for minimizing air movement in the optics shop, and a dozen other ideas that touched the project. When Porter visited the West Lynn laboratories, Ellis regaled him with those ideas, and with reports on their work on perfecting fire control on warships. The one thing Ellis didn’t offer was a date when a usable disk would be ready.

  In Pasadena, Pease’s design work waited for progress on the mirrors. He turned his attention to the still-troublesome one-hundred-inch telescope on Mount Wilson. Hubble and Humason were using the telescope regularly, as were other astronomers, and on good nights it produced excellent results. But the falloff of the image quality in some areas of the sky was still disturbing, and the apparent inability of the mirror to settle down, after even modest changes in the temperature, remained troublesome.

  Pease had worked out some ideas for edge mountings for the mirror of the two-hundred-inch telescope. He got permission to use the one-hundred-inch as a laboratory to test his ideas. What was happening, he concluded, was that the mirror of the one-hundred-inch telescope was changing shape, deforming from its own weight, as the telescope was tilted in various positions. The problem hadn’t caused great concern on earlier telescopes because it was thought that the massive solid disks in telescopes like the sixty-inch and the one-hundred-inch would minimize the deformation. But the demands of a telescope increase with the size. A distortion of one-tenth of a wavelength of light—a distance measured in millionths of an inch—may be difficult to measure on a small telescope; it becomes readily apparent in the image quality on a larger telescope. Pease worked during the daytime and during light-sky portions of the month, when the moon was up and the sky was too light for deep-sky observations of remote galaxies.

  Ellis’s spray process for the sixty-inch mirror transformed the working area inside the steel building at West Lynn into a self-enclosed hell on earth. A bank of transformers along one wall hummed with the eight hundred kilowatts of power needed to heat the twelve-foot-diameter furnace. The steel walls reverberated with a low-frequency hum that made talk impossible. The temperature inside the furnace was over 1900°F, and heat waves caused the air in the building to undulate until the walls and equipment seemed to shimmer. The temperature at the nozzles of the burner was double that of the disk itself, close to 4000°F, so hot that the burner had to be shielded in an enclosure of fused quartz to withstand the temperature. A heavy outer pipe shielded the smaller pipes for quartz powder, hydrogen, oxygen, and cooling water to the burner. Technicians could only view the process through a thick, green glass window set into a steel viewing protector.

  In the first trials the new burner apparatus worked exactly as Ellis had planned, laying down fused quartz at a much higher rate than even his optimistic predictions. The intense sleet storm of quartz that fell into the mold surface appeared to fuse readily with the material laid down behind it, and the disk built up quickly. It wasn’t until the trial disks cooled that Ellis discovered that the layers of quartz had failed to fuse. Instead of a single, fused disk, what emerged was a lamination of partially fused layers, a quartz millefeuille pastry.

  Ellis’s answer was more tests. Each cycle of heating and cooling the furnace took at least a week and an incredible quantity of fuel, so Ellis went back to the small furnace for testing. The only way he could get the quartz to fuse completely was to slow down the rate of spraying until it was too slow to be practical. More tests finally identified the problem: The new process for pulverizing and filtering the quartz filtered out exactly those fine particles that would have provided the fusion between the layers of sprayed quartz.

  Ellis decided to start over again and prepare new batches of quartz instead of reprocessing the quartz that had already been prepared. He would use cheap native quartz for the body of the disk, instead of expensive Brazilian quartz. The native quartz had to be hand selected to avoid white streaks, which seemed to produce bubbles. Ellis assigned three men to the job of sorting an enormous rock pile of quartz into three smaller piles of quartz with no white streaks for the faces of disks, pieces with minimal white streaks for the base material, and a discard pile with excessive white streaks. Seventy-five percent of the quartz fell into the first two categories and was then ground in the ball mill.

  Ellis tried the newly prepared quartz in another experimental disk. If the quartz had even minimal white streaks, disks came out looking “more like porcelain than quartz because of the great number of small bubbles.” When he tried to fuse a layer of pure Brazilian quartz onto the bubble-laden base, the heat transferred to the base during the process expanded the bubbles, raising the mass “like yeast raises bread.”

  Guessing that the problem might be moisture in the quartz, Ellis tried drying samples for periods from one hour to one week, at a temperature of 450°C. The drying made no difference. No matter what he tried, he could not produce a satisfactory disk. He kept turning back to an earlier experimental disk, number thirteen, made by the older procedure, from the identical quartz, from the same quarry, even from the same part of the quarry as the material they were currently using. That blank had come out essentially bubble-free. When he tried a sample of quartz left over from the earlier disk, it fused beautifully. It was as if he had left out a magic ingredient that had made the earlier tests work.

  Hands-on experimentation was the specialty of the West Lynn laboratory, the core of Thomson’s reputation. Ellis had notes from the various stages of their experiments, but his methodology was trial and error, adding more or less of one ingredient or another, or substituting one grade of quartz for another. No one at GE fully understood the processes at work inside the furnaces. When Ellis offered a description of the problems, it wasn’t the sort of rigorous explanation scientists like Anderson or Hale would expect:

  The pulverized quartz does not become melted while passing through the flame from the burner to the work, but is caught in the sticky surface of the work and melted there. Some of the quartz is vaporized from the surface and perhaps from some of the extremely fine material passing through the burner. This vapor condenses on the cooler parts of the surface and furnace forming a white spongy layer of varying thickness. To this deposit is added some of the larger particles of pulverized quartz by the action of the flame, or by pieces bounding from surfaces that have not yet reached the sticky state. The surface layer thus formed seems to be a very good heat insulator, and which must be melted by heat transferred through the layer of quartz being laid down by the burner as it passes over the surface.

  Ellis concluded that even a trace of “white material” in the quartz would contribute excessive bubbles or contaminants of iron or chromium to the laid-down quartz layers. The bubble-filled material then became so effective as an insulator that it did not pass the heat to the layers beneath, preventing complete fusion of the layers.

  Desperate to get the process working again, Ellis tried alternate materials, like flint shot, a pure silica sand from an enormous deposit in Ottawa, Illinois. He had used the sand successfully when he molded the bases of the earlier disks, but when he tried pulverizing a new batch for the sprayer, the quality and purity varied so much that a process to prepare production quantities would require “more development.” He finally went back to the quartz pile in the yard outside the laboratory. For the next batch Ellis had a single man grade the quartz in an effort to assure uniformity of the grading standard. The man would study each piece of quartz against a black background, holding the samples underwater to minimiz
e reflections that might hide the white streaks. A quick test of the new quartz appeared to fuse, and Ellis hoped that by December 1930 he would have enough quartz on hand to begin the long-awaited sixty-inch mirror blank.

  At least, Ellis assured Hale and Anderson, the current problems were the last they would encounter. “Every man on the work is doing everything he can to produce a 60-inch mirror at the earliest possible date, for with this experience behind us the rest will be easy. There will, of course, be problems to be solved in making the 200-inch mirror, but it does seem to us that practically everything that can happen has already happened, and we have every hope that the coming year will be brighter.”

  He finally started spraying a sixty-inch mirror for the telescope on December 8, 1930. Miraculously, everything worked. On January 6, 1931, Ellis triumphantly telegraphed Hale: “We have laid one more ghost. The first sixty inch mirror blank has been reduced from annealing temperature approximately 1100 degrees C. to room temperature in eight days, an astoundingly short time compared with glass.” Finally, it seemed, Ellis had licked the fused-quartz demons. Hammond in the GE publicity department prepared another article for the journal The Glass Industry, celebrating the achievement. A GE executive named McManus, at the main office in Schenectady, began negotiating with representatives of other observatories, including Harlow Shapley at Harvard, for future orders.

  In Pasadena, Hale and his colleagues were delirious with excitement. Walter Adams and Theodore Dunham of the Mount Wilson staff traveled to West Lynn to see the sixty-inch when it came out of the annealing oven. They liked what they saw. Even Hale was encouraged. “If it were not for the ‘fierce’ cost,” he wrote, “I should feel much encouraged. Their capacity for spending is appalling…. although they are two years behind their original time schedule some of us may live long enough to see a 200-inch disk, if the money holds out.”