VI—
American Cyclotronics

Up to 1935 the 27-inch cyclotron ran in fits and starts. Then a series of improvements increased not only reliability but also energy and current: the deuteron beam rose rapidly, one µA or two at 3 MeV in 1934, which "thoroughly sold [initiates] on the cyclotron as the perfect high voltage source," to twenty µA at 6.2 MeV early in 1936, which established the machine as the preeminent producer of radioelements in the world.^[1] The enlargement of the poles to 37 inches, which occurred during August of 1937, brought a beam of about twelve µA at 8 MeV, which in the following months grew to seventy-four µA, and to twice that at 6.2 MeV, the old maximum energy of the 27-inch cyclotron. The 60-inch, operational in the summer of 1939, was running in 1940 at new levels of energy, current, and dependability. Most of the instrumental changes that created the bigger beams were invented in Berkeley. In the case of two major design features of the 60-inch, however, the Laboratory adapted what other cyclotron laboratories in the United States pioneered.

1—
The Art

Berkeley's Poulsen Machines

Figure 6.1 shows the rise of beam current and energy of the cyclotrons built under Berkeley's Poulsen magnet from the onset

[Full Size]

Fig. 6.1
Vital statistics of the Laboratory's 27-inch and 37-inch cyclotrons. The ordinate
indicates deuteron energy in MeV; the points are labelled with currents in µA.

of substantial improvements to the start up of the 60-inch. The figure discloses that the advance occurred in five steps: (1) an increase in beam from June to October 1935; (2) a doubling of current and an increase of energy of 150 percent between November 1935 and March 1936; (3) high currents at relatively low energies, reached in the summer of 1936; (4) achievement of steadier operation; (5) a jump in the energy maximum and a doubling of current at the previous maximum between August 1937 and February 1938. The causes of these advances will indicate the range of considerations and problems the cyclotroneer had to heed.

Step 1 provides a notable instance of the principle that whatever can go wrong will. In the spring of 1935 Lawrence urged his boys to create larger deuteron beams at higher energy in order to make enough radiosodium for clinical work. His goal was modest: 5 µA at 5 MeV or 50 µA at 3.5 MeV; his eventual goal, as he told

Cockcroft, an immoderate milliampere. The boys tried many ways to a bigger beam: widening the dees, enlarging the filament of the ion source, putting up the dee voltage, and so on, all "with disappointing results." The summer was a "depressing time," "an epidemic of trouble;" whenever conditions augmented the beam, cathode rays somehow generated in the cyclotron tank would penetrate the glass insulators supporting the dees and shut down operations. It was very frustrating, this attainment of as much as 9 µA followed almost immediately by a crash. Lawrence wanted 10 mCi of radiosodium a day, which required a steady minimum current of 10 µA. "I was almost driven to distraction."^[2]

He was not so distracted that he neglected to reassure his backers. "You will be interested to know," he wrote Poillon toward the end of the frustrating summer, "that we have been quite successful already in stepping up the production of radio-sodium to the point where it is now practical to begin clinical examinations." Or would be, if the machine could sustain the currents produced. "It was planned to start treating a patient at the University Hospital last Sunday with radio-sodium, but unfortunately the cyclotron broke down." The boys were working on it. "I am sure before long we will have improved the construction so that the apparatus will withstand the higher power."^[3] He was right. During September supersleuthing traced the problem to a change in conductivity of the water cooling the vacuum chamber.

From the beginning, apparently, an unknown and unnoticed rectifying action had tended to establish a dc potential on the dees. Previously the water had been sufficiently conducting to prevent the accumulation of any significant charge; in the summer of 1935 it no longer carried the charge away. Following this subtle diagnosis ("Who in the world would have guessed it?"), the cure was obvious: a choke coil introduced between the accelerating system and ground destroyed the dc potential on the dees, the cathode rays stopped, and all troubles vanished immediately. That was around October 1, 1935.^[4] Within a few weeks the

cyclotron gave 10 µA steadily at 5 MeV, all the target could take. "We can almost detect the neutron ionization here in Princeton," White wrote in admiration. The fix was durable. In the summer, with its constant shortings, the cyclotron gave "an injustly unfavorable impression of its performance;" at the end of the year it was "practically robotized" and almost as likely to work as not.^[5] Still it was by no means dependable, and Lawrence could not take advantage of an offer from the Macy Foundation, engineered by Poillon, to give John Lawrence and the Mayo Clinic each $1,000 for the purchase of radiosodium for biomedical research. "We are not in a position to supply radioactive substances for clinical work. . . . Our apparatus runs only spasmodically, and a good share of the time we have it dismantled for repair and alterations."^[6]

This brings us to step 2, the increase of energy to 6.2 MeV and of current to 20 µA, obtained in January and February 1936. It was primarily Cooksey's doing. He made a new vacuum chamber that provided for, among other things, a larger final radius for the spiralling ions and many small adjustments suggested by two years' experience with the machine. Cooksey's chamber did without the notches that Lawrence had cut into the old one to alter the magnetic field in the desperate days of the previous summer. The new chamber worked almost immediately.^[7] This triumph of cyclotroneering—a major intended alteration that required no fiddling, an important step toward realizing "an apparatus engineered to the point where we can depend on reliable operation to an extent justifying beginning clinical research"—may have been decisive in causing Columbia and Cambridge to proceed with the construction of cyclotrons.^[8] In mid

March the 6 MeV beam came out of Cooksey's chamber through a platinum window ten-thousandths of an inch thick. It made a splendid spectacle, a cylinder of light over ten inches long streaking through the air. This accomplishment had more to recommend it than mere display: it made possible experiments beyond the influence of the Poulsen magnet, thereby cancelling (as the Illinois cyclotroneers put it) "one of the formerly objectionable features of the cyclotron."^[9]

Cooksey's can was to provide "for intense neutron sources over long periods of time and for large quantities of radioactive substances." By doing so, it showed the way to more of both; and in the summer of 1936 another chamber, with dees widened to two inches (from 1.25 inches) and a larger filament, replaced the one installed in January. This brings us to step 3: the steady current more than doubled to 50 µA, albeit at 4.3 MeV, and the peak reached over 90 µA, with no end in sight. It was at this time that (in Lewis's words) the neutron flux became a "public nuisance," disturbing experiments in the Chemistry Department 300 feet away; Lawrence reported to Poillon, with fatherly pride, the words of Mayo's Alvarez, "our baby has become a monster." Lawrence set his people to making 50 µA at the maximum energy of 6.2 MeV.^[10] In December 1936 they almost complied. "For some reason, for which we are not entirely clear," the current shot up to 35 µA at just under 6 MeV with the 1.25-inch dees in place. Cooksey guessed that changing the current to the filament from dc to ac might have done it (with ac the filament could operate without bending in the magnetic field). Odder yet, the big current came out without any shimming. "This delightful mystery is unexplained. One might say unexplained either way you look at it, for no one knows why we had to use shims, or why we do not have to use shims." "There are certainly lots of things to learn about the

cyclotron." With a little shimming and four-inch dees, it gave more than 150 µA steadily at 5.5 MeV, and 170 µA peak.^[11]

In step 4, which occupied much of the first half of 1937, effort went to improving reliability of performance. There were two matters identified in 1936 that particularly needed attention, one simple and mechanical, the other complicated and electrical. The first was to do without wax. "If only we could get rid of wax joints," Lawrence sighed, "I am sure the outfit would run with very nice regularity."^[12] Cooksey set to work on another chamber, which was "to be a humdinger," with so large a vacuum gap (some 7.5 inches) as to allow for dees three inches wide and very thick, and with rubber gaskets in place of wax seals. (The large gap left enough space between the wide, thick dees and the chamber walls to keep the capacitance of the system small; the thick dees resisted buckling, and hence change in capacitance, during operation.) Lawrence anticipated, rightly, that Cooksey's gaskets would be a "tremendous improvement," a great step forward in convenience and reliability.^[13] The electrical matter concerned that perennial irritation, the high-frequency oscillator, which by January 1937 was on its last legs.

This oscillator worked according to Livingston's old plan, with a self-excited, tuned-grid, tuned-plate circuit. Its heart was the power tube developed by Sloan and improved with the help of Thornton and F.A. Jenkins. This water-cooled cylinder, rated at 30 kW, had a copper pipe as anode, copper wire wrapped on copper poles as a grid, and a long tungsten wire bent into a hairpin as a filament. When working it was continuously exhausted by a pump working with Apiezon oils. The Laboratory made these tubes itself, or had them made in the shop of the Physics Department, which brought their cost down to perhaps two-thirds

that of the equivalent commercial models.^[14] The oscillator employing them had the advantage of simplicity and robustness. But it also had a serious defect.

The Livingston-Sloan oscillator drifted in frequency. To compensate the operator had constantly to adjust the magnetic field, by inspiration and experience more than by science, and thought himself successful if he could keep the beam sturdy and steady to within 10 percent. The current exciting the magnet could easily have been kept constant to within 1 percent with a feedback method, or automaton, of a type developed at the Cavendish; but such a device would only have hampered the fiddling necessitated by the drifting oscillator. There was something else: the oscillator circuit had a tendency to parasitic vibrations at unwanted frequencies, which uselessly consumed power wanted on the dees. This loss, added to consumption within the tube, reduced the effective rf power on the dees to half that of the oscillator: 12 kW to accelerate ions out of 25 kW purchased from PG&E. As Cooksey wrote Cockcroft, who had begun work on a Cambridge cyclotron: "Our methods of operation have been far from ideal and we know it."^[15] To idealize them Lawrence engaged Charles Litton as the technical assistant provided for in the charter that had established the Laboratory. Litton had been the chief engineer at Federal Telegraph; he constructed in accordance with the production standards of high-tech industry, not in the experimental and jerryrigged fashion of a physics laboratory. A 1-kW radio transmitter of tightly controlled frequency regulated the output from two 20-kW tubes. Lawrence watched in admiration. "It [Litton's oscillator] has been designed and built in the regular style of the first class engineering job that one sees in a broadcast station. It should make the operation of the cyclotron simply a matter of pushing buttons."^[16]

Litton's "de luxe oscillator," as Lawrence called it, was installed in March 1937. "You can imagine the joy in tearing down the old oscillator and its associated junk." The de luxe way worked. "We now control the magnet with an automaton [Cooksey wrote the cyclotroneers at Rochester] and the beam remains constant to within ten percent for hours without attention. The boys are all complaining because the cyclotron has become so dull."^[17] This is not to say that the age of heroism had passed. In late May or June the oscillator went down, leaking; the trouble was traced to the pump, and Litton summoned from Redwood City, where he had set up in business. A replacement pump arrived at 2:00 A.M. on a Sunday morning; Richardson, Oldenberg, and Van Voorhis labored day and night to revive the oscillator, but the beam would not rise above a miserable 2 µA. This time the trouble was not in the oscillator. Alvarez had placed a new cloud chamber between the cyclotron's pole faces. The iron in his equipment adversely shimmed the field. When it was put aside, the cyclotron recovered the beam and the reliability that the Laboratory had come to expect.^[18]

We come to step 5. Alvarez's experiments were the last to be done on the 27-inch cycloron. During the first few weeks of August 1937 the Laboratory achieved a goal that Lawrence had been considering since February 1932 and planning since the spring of 1933, when he wrote Poillon of his desire to enlarge the poles of the Poulsen magnet to make possible the acceleration of particles to 10 MeV.^[19] The $1,550 required had been available neither then nor in the summer of 1934, and subsequently improvement of the 27-inch had claimed priority. In January 1937, however, with the prospect that Cooksey's gaskets and Litton's oscillator would "approximate our dreams of a satisfactory outfit for nuclear work," it was decided to enlarge the poles to 37 inches during the summer. The goal was not only to increase the beam, as usual, but also to test a new chamber on Cooksey's

new principles for scale-up for the Crocker cyclotron. If the chamber, made from a ring of cast brass and sealed with gaskets, proved itself, Lawrence would have "no worries about funds for the new lab; but if it doesn't give fairly high output we will have to find out why as soon as possible."^[20]

The new pole pieces were ordered from Moore Dry Dock Company early in June 1937 and delivered, for $490, a month later. On July 9 Cooksey's 37-inch chamber was moved into the Laboratory by block and tackle anchored by Cooksey's Packard. It leaked, but clammed up completely after liberal application of glyptol, a thick enamel for insulating wires.^[21] On August 4 the 27-inch concluded its pioneering life. In eight hours the University's maintenance department installed the new pole pieces. (The business was not altogether trivial; the corresponding department at Ohio State dropped a pole during installation there and decreased its parallelism by a factor of three.) A few days later the magnet gave 14 kG. Around August 15 tests began, with new pumps, new water cooling, and new circuit components.^[22] The machine was expected to run immediately on switching on and almost did so, with a little shimming by Kurie. By the end of August there was a very big beam: 100 µA at 4.6 MeV, steady, good for more than a curie of radiosodium. "Thrilling results," said Poillon.

Cooksey's new tank ran for a month without springing a leak, which, in comparison with earlier chambers, his included, amounted almost to a perpetual hermetic seal.^[23] Litton's handiwork did not do so well. In order to get high currents near maximum energy, more power was required on the dees than the de luxe oscillator could deliver. The Laboratory went back to a homemade model using pumps three times as fast as Litton's.^[24] A

competition to improve the creation of nothing arose between Litton and Sloan. A better pump resulted, with a new fractionating system and Litton oils rated by the inventor over Apiezon products. An estrangement resulted too. Litton thought that the Laboratory made too free with his ideas and injured his chances to patent them. It was a natural consequence of the communal ethic of the Laboratory and the vacuous patent policy of the University.^[25]

During the first few months of operation of the 37-inch cyclotron, Robert Wilson was engaged in a lengthy analysis of the paths of ions between the dees. In January, shimmed to his specifications to increase the field near the center, the machine gave 170 µA at 5.5 MeV, and, a few weeks later, 150 µA at 6.2 MeV. The sensitivity of the beam to shimming may be illustrated by data collected by McMillan: Wilson's wonderworking pyramidal shims, made of ten iron disks 0.013 inches thick and 6.5 to 20 inches in diameter, cut the beam from 6 µA to 0.5 µA when placed above the chamber, and restored the 6 µA when placed below; two such pyramids, one on top and one on the bottom, made 28 µA; removal of two disks from the upper stack increased the current to 30 µA. Pyramids are full of mystery. They did not work at all at Rochester, as its chief cyclotroneer told Lawrence, in bewildered admiration of what Berkeley had achieved. "Each new figure seems more unbelievable than the last, and we wonder if there is going to be any limit to what you can extract from your machine!" All this exceeded the target's, if not the cyclotron's capacity, and the Laboratory tended to run at 100 µA or less for isotope manufacture.^[26]

It also ran with unprecedented regularity, "almost night and day," as Lawrence wrote early in January. In the six months ending in September 1938, the 37-inch cyclotron operated for almost

eight hours a day on average, seven days a week, and sometimes twenty-four hours a day in three eight-hour shifts. For seven or eight weeks at the end of the year, during the sojourn of an aspiring Belgian cyclotroneer, Paul Capron, the machine worked "without any serious trouble" from 8:00 A.M. to 11:00 P.M.; it attained a peak current of 300 µA of 8 MeV deuterons and made neutrons as plentifully as might 10 or 100 kilograms of radium. "In Berkeley I realized exactly all the possibilities of the cyclotron." Rather than apologize for the erratic performance of their machine, cyclotroneers could now consider it "a piece of standard equipment, a laboratory tool which can be put up by [Berkeley] experts in a short time, at reasonable expense, and without undue local development work." The weekly schedule of oilings, greasings, and other checks, introduced by Brobeck in the summer of 1938, helped to keep the machine regular.^[27]

The bottom line, however, was not regularity of operations but quantity and efficiency of production. The initial routine for making P³² , the most readily manufactured isotope potentially useful in medicine, will indicate the process and the possibilities for improvement. One began with red phosphorus, a very disagreeable substance, which, in the system developed by Kurie, was rubbed into grooves in a water-cooled copper plate maintained in an atmosphere of helium. This elaborate target received deuterons in the usual position, outside the dees, and spread radiophosphorus everywhere. At 8 MeV it took 35 µA-hours to make 1 mCi of P³² . John Lawrence had set 1,520 mCi as the amount needed for experimental therapy. In the spring of 1938, when the 37-inch cyclotron delivered 60 µA of deuterons steadily at 8 MeV, it had to labor a day and a night to make one clinical dose. The case was worse for Fe⁵⁹ , which, because it derived from the rare iron isotope Fe⁵⁸ , could scarcely be made in quantities sufficient for tracer work—some 1 µCi per 100 µA-hr.^[28]

Some creative thought about the electrical bill indicated a route

to important improvement. Experience showed that about half of the power expended in the dee circuit went to drive the ions. Hence in normal operation as much as 10 kW must have been tied up in the energy of the particles circulating within the dees. But less than a single kilowatt of beam heated the target. Wilson looked where reason pointed; toward the end of April 1938, he found that the deuteron current inside the dees was ten times what emerged from them.^[29] In June, with 50 µA on the target, Wilson's probes detected 650 µA at about two-thirds of the distance from the center to the deflector. Since the presence of the probe did not diminish the target current, it appeared that the Laboratory could bake material for biologists while giving an external beam for physicists. It was a godsend. "One cyclotron may satisfy the needs of both biologists and physicists without extremely long periods of operation."^[30]

For universal satisfaction, the raw material for the baking has to be available in a refractory form. Ferrous phosphide answered perfectly. By covering probes with it, cyclotroneers could make P³² and Fe⁵⁹ simultaneously, much more quickly and with much higher specific activities than by Kurie's method. In developing this technique Wilson collaborated with Martin Kamen, a Ph.D. chemist from Chicago, who had arrived around Christmas 1936 in order to work with Kurie for nothing; he had ascended to salaried research assistant the following July, when Lawrence learned that he was capable of overseeing the preparation and distribution of the cyclotron's products. Wilson and Kamen's innovation raised yields by an order of magnitude. It was of the first importance for the Laboratory's functioning and financing. Kamen remembered almost fifty years later: "Bob and I were the heroes of the moment."^[31]

We have operating details from the end of August to the middle of November 1939. The machine ran on most days for an average

of eight or ten hours; all of its internal and about 60 percent of its external bombardments went for preparation of isotopes for biomedical work. Otherwise it stopped for improvements, for example, the replacement of the old pyrex dee insulators by new Corning ware; for lack of business, as the 60-inch came on line; and for the inevitable repairs. Trouble occurred in the usual places for the usual causes: "filament burned out for no apparent reason," "oscillator failed to oscillate, fixed itself," leaks here and there, a cracked target window, blown fuses. An adjustment to the position of one dee opened holes in the chamber seal that took a day to locate and fix; then the oscillator would not oscillate; and so on.^[32] There were other times when the 37-inch cyclotron acted erratically or quit altogether, and in 1940 it shut down for at least two extended periods for rewiring the controls and for rebuilding the oscillator. No doubt its later fitfulness, as compared with its earlier dependability, arose from its feeling of neglect. "If anyone cares about the old 37-inch cyclotron," Aebersold wrote, in announcing a restoration to service in 1940, "it is running fine again."^[33] From early in 1938 the Laboratory had been preoccupied with assembling, testing, and operating the 60-inch cyclotron.

The Crocker Cracker

"It is a truly colossal machine." Thus Edoardo Amaldi, who passed through Berkeley in September 1939. The cyclotroneers in partibus, who knew it first from Cooksey's photographs, gasped at the size: "almost unbelievable" (Lyman); "almost too much" (Van Voorhis). Figure 6.2 shows what caused "widespread admiration and awe," what made "everybody . . . open mouthed" among the cognoscenti.^[34] The giant's magnet weighed 220 tons; it stood 11

[Full Size]

Fig. 6.2
The Crocker Cracker as portrayed in the University's student publication,
The Pelican , 1939. LBL.

feet high; it consumed 45 kW of power. Its frame could hold the entire staff of the Laboratory (see plate 5.2). Barnes compared the ports in the vacuum chamber to manholes; Laslett admired the "amazing 153 cm beam;" Snell joked that its neutrons would reach Chicago; Aebersold worried that five feet of water did not diminish radiation enough for safety in Berkeley and that all the cyclotron's targets would be too hot to handle.^[35]

Why was the Crocker cyclotron made so big? During the planning stage, Lawrence wrote Cockcroft that "the new cyclotron will be much larger than needed for any work contemplated in the near future." The rectangular design of the magnet was wasteful of material; as the American Rolling Mill Company (ARMCO), who bid on its construction, pointed out, the large spaces between the poles and the frame diverted much of the magnetic flux from the 22-inch gap containing the chamber. Lawrence acknowledged the inefficiency of the design, which he explained to the ARMCO engineers, as he later did to Chadwick: the 220 tons of metal in the magnet exceeded what was needed for the intended voltage and current "because it has been designed for the primary purposes of medical research, requiring openness and accessibility." Maximizing the flux and field between the poles had not been the goal; as usual, the Laboratory sacrificed efficiency to allowance for the possibility of future change.^[36] And yet, as Lawrence made clear to a colleague wishing to make a medical cyclotron, "We recognized that for medical purposes alone such a large installation was hardly justified. On the other hand, for the sake of physics, it was important to get up to as high energies as possible." But in that case, a more efficient design would have been justified. It appears that no good reason or clear purpose except flexibility fixed the dimensions of the 60-inch cyclotron. Why then was the

Crocker Cracker so big? "Since we can get the money for [it]."^[37]

"The cyclotron has now passed beyond the stage of ordinary laboratory development; its further improvement involves problems of a primarily technological engineering character." So Lawrence wrote in his report to the Research Corporation for 1937. The builders of the 60-inch cyclotron would enjoy not only a sufficiency of funds but also the luxury of engineering from scratch. No hand-me-down magnets, no industrial discards, in short, as Nahmias put it, "no patch work." Everything had to be planned in advance: the inefficient magnet designed and tested on a scale model, bids sought and tendered in accordance with detailed blueprints, tons of metal machined and installed. The magnet and its building had to be designed to resist earthquakes. "The prospect of this job is rather appalling," Cooksey wrote shortly after starting it, "as it really seems to be engineering rather than experimentation."^[38] A poignant example: Cooksey found the mold in which the 60-inch chamber was to be cast "so appalling in its complexity" that he could not understand how his own design was to be realized.^[39] The Laboratory needed a first-class engineer. To its extraordinary good fortune, it got just the man it required, just when wanted, and for just what Lawrence liked to pay. William Brobeck came to work for nothing in the fall of 1937; by March 1938 he had displaced his friend Cooksey as chief cyclotroneer. "[He] puts me in the shade."^[40] Brobeck reworked the original sketches, made when funding and prudence recommended 50-inch poles, into engineering plans for a 60-inch magnet based on designs by Alvarez.^[41]

The Crocker Laboratory came into existence slowly because the architect of the building eluded Lawrence's pressure by dying.

The delay in finding a successor left a long time for dickering for copper and steel, the two largest single items of expense. Lawrence tried to get it gratis through Buffum.^[42] He did not obtain even a price concession; on the contrary, the price of steel rose and the magnet climbed over budget. Columbia Steel, a San Francisco firm, obtained the plates and disks for the frame and poles from Carnegie-Illinois Steel; Columbia's management took an interest in Berkeley's "atom buster," as they called it, and kept the price of the steel under $12,300 f.o.b. in box cars in San Francisco. Machining, by Pelton, cost $3,000. The order was placed at the end of March 1937; twelve months later the 150-ton yoke stood in place in the Crocker Laboratory.^[43] The metal for the windings and the coil tanks came to about $400 less than the steel; Revere Copper and Brass supplied the copper in strips and a carefully considered plan, "rather difficult . . . to delineate," for winding and insulating it. The copper and the plan were turned over to a local firm, Gardner Electric, who underbid Revere for the construction of the coils. Gardner discovered that Revere had furnished some strip so rough as to threaten the paper insulation. With Gardner's vigilance, the Laboratory received carefree coils, which had taken their place under the yoke by early June 1938. The total magnet was complete in August.^[44]

The cost of the Crocker magnet, including assembly, came to $31,546, very little more than the value assigned to the much smaller Poulsen magnet in 1930. The reduction in unit cost was a consequence of laminating, rather than casting, the frame and the poles, a method that ran a risk of lopsidedness. "It is a very satisfying achievement," Cooksey wrote, after determining that the poles came parallel to within 0.0014 inch, "to have achieved such parallelism out of what one might almost consider a stack of

cards. It certainly shows our colleagues in other laboratories an inexpensive way to build a magnet for cyclotron use."^[45]

Entrusting orders to local firms, which proved itself in the procurement of the steel and the winding of the coil, also brought an excellent product in the tricky matter of the vacuum chamber. Cyclotroneers disputed the relative disadvantages of rolling and casting the brass ring that made up the chamber walls. Cooksey favored a casting for flexibility, knowing, however, that it might be too porous to hold a vacuum. Moore Dry Dock Company of Oakland undertook to try. They poured 2,838 pounds of pure brass into the mold whose complexity befuddled Cooksey; and they made a casting full of holes. A second try produced a chamber vacuum tight but for one spot, which was plugged up satisfactorily.^[46]

While awaiting the delivery of the magnet, the Laboratory organized itself into a number of interlocking groups to design and build the accelerating system and controls of the big machine. The scheme, as proposed by Brobeck in November 1937, reads like an industrial, or even military, table of organization: "The work of building each group of apparatus is given to a committee for which a leader is appointed. . . . He may divide the work as he wishes. . . . Supervisors are to be appointed for definite lines of the work and their approval must be obtained by the committee leaders for all work included in these lines. Committee leaders, supervisors and others needed are to meet periodically with the directors of the Laboratory . . . [as] the directing committee ." Examples of committee assignments: magnet, vacuum chamber, vacuum pumps, oscillator. Examples of lines: mechanical design, vacuum systems, wiring, radiation. Brobeck provided for eight committees in all, as did Cooksey in a memorandum written two weeks later; their discordant schemes drove a finer division of labor, which ended in a definitive score in the summer of 1938.^[47] The division is depicted in table 6.1.

During the summer of 1938 the "directing committee"—initially composed of Lawrence, Cooksey, McMillan, Alvarez, Snell, and Brobeck—met at least three times at the Leamington Hotel in Oakland to decide policy at a decent distance from those who would carry it out. They constituted themselves supervisors, as in Brobeck's original proposal, in accordance with their expertise: Cooksey took charge of the chamber and ion source; McMillan, low-voltage power and wiring; Alvarez, radiation protection and electronics; Snell, controls and instruments; Brobeck, the magnet and mechanics. John Lawrence was elevated to the directorate and given all responsibility for medical matters; Winfield Salisbury remained a mere committeeman, but received authority over the oscillator and its power supply. To coordinate operation of the 37-inch cyclotron with the completion of the 60-inch, the directing committee decided to choose crew captains exclusively from their number.^[48]

Snell's leaving for a position at Chicago (his assignments fell to McMillan) further restricted the pool of captains. The directors deliberated whether to enlarge it by promoting midshipmen Simmons, Abelson, and Aebersold, but decided against the dilution. The matter concerned safety as well as efficiency. Lawrence worried that the crews did not observe routine safety practices, knew nothing about first aid, and exposed themselves recklessly to electric shock. Salisbury received a nasty shock from the emission voltage of the filament, a potentially most serious mishap, which could have been prevented had he used the safety switch provided. Strong warnings were required, and sets of instructions, should time be found (none was) to write them down. Meanwhile only Alvarez, McMillan, and Brobeck would enjoy the rank of captain. Their crews would work in three watches, 8–12, 1–6, 7:30–11:30. To complete bureaucratic arrangements, Cooksey was commissioned to put signs on all the external doors of the Laboratory: "No admittance without appointment."^[49]

It would not be profitable to follow in detail what went on behind these closed doors. But one feature of the development of the 60-inch deserves emphasis: the construction of cyclotrons elsewhere had eased the burden of innovation in Berkeley. In two important matters—the ion source and the dee system—the Laboratory adapted what others had introduced. In the 11-inch cyclotron, ions formed in the field of an exposed cathode, which perforce operated at the residual pressure within the "vacuum" chamber. The source for the 27-inch was only slightly more elaborate: a coiled filament insulated by glass sheathed in copper sat above (or below) the dees at the center of the diametral gap and sent its stream of electrons along the magnetic lines of force into the bottom (or lid) of the cyclotron chamber. Livingston took steps to improve the method after his translation to Cornell. He began by adapting the capillary fountain designed by Tuve and his associates for use in their high-tension apparatus. They placed an anode and a cathode opposite one another along the axis of a metal tube, narrowed the space between them into a "capillary" in which the ions formed or collected, and drew the ions out perpendicularly to the axis through a small hole. By rapid pumping outside the capillary, they could maintain the gas within it at a pressure 500 times greater than that in their accelerating tube. They managed to tease out of the hole a positive current of no less than 1,500 µA.^[50]

Livingston's variation of Tuve's capillary yielded a current of positive ions of about 500 µA in December 1936. Of these perhaps a fifth were protons; it remained only to conduct them from the capillary hole to the collection cup or target. Collecting proved taxing and at first Livingston and his colleagues at Princeton and Purdue, who tried to follow his lead, got but a feeble current. Profiting from this informal testing service, Lawrence and Cooksey stayed with what they knew to work in the conversion to the 37-inch cyclotron, although Cooksey did not like to recommend their flat filament to others. Lengthy tinkering—paid

for by the Research Corporation—eventually multiplied the deuteron beam at Cornell twelvefold, from 3 to 35 µA, and then doubled that. In September 1938 the Laboratory was planning a big capillary. "It should give 4 m.a.," Cooksey wrote Lawrence, "Oh dear." Livingston's version of 1938 is shown in figure 6.3: from the capillary's hole at the cyclotron's center issues a stream of positive ions created in the plasma around the hole by electrons shot from the hot cathode at the top of the source. The gas to be ionized enters the upper alembic-shaped vessel through the pipe that also brings current to the filament. The little circles on the surface of the alembics indicate cooling tubes. The hooks or "feelers" extending from the left dee toward the hole supply the electric force to extract the positive ions. With this arrangement, Livinston and his associates brought 70 percent of the 100 µA of protons issuing from their source to the collector and the cyclo-

[Full Size]

Fig. 6.3
Livingston's capillary source. Livingston,
Holloway, and Baker, RSI, 10 , (1939), 65.

troneers at Purdue managed to obtain five or six times the current to target they had from a hot tungsten filament.^[51]

As Livingston observed, his handiwork had important advantages in addition to increasing the current. The beam was not only bigger but also narrower and better focused owing to the small size of the exit hole and its placement at the center of the cyclotron. At Berkeley the circulating current was ten times as large as the beam to target, whereas Livingston brought home 70 percent of the protons he started with. Because of the better definition of the beam, the wide dees in use at Berkeley to accommodate the ions from the extended exposed filament were no longer required. The vacuum and magnet gaps could be narrowed, reducing power consumption and increasing the field and therewith the maximum available energy.^[52] When McMillan and Salisbury finished their other chores for the 60-inch cyclotron, they adapted Livingston's paragon in the manner shown in figure 6.4. Here a truncated metal cone, which Livingston had mentioned as a possibility, serves as anode to a hot cathode at its base. The exit hole at the top of the cone occupies a place just off center in the cyclotron's median plane. After some experimentation with the size and orientation of the hole and the feelers, McMillan and Salisbury obtained 90 µA of deuterons to the target. Under the same conditions, the old filament source delivered 14 µA.^[53] To take advantage of the reduction in the vacuum gap made possible by the capillary source, one put iron disks ("filler plates") into the chamber or, as at the Carnegie Institution, made them integral with thick cover plates.^[54]

[Full Size]

Fig. 6.4
The Laboratory's modification of
Livingston's capillary source. McMillan
and Salisbury, PR, 56  (1939), 836.

The old dogs learned new tricks also in the design of the dee system. In the standard Berkeley arrangement, which most early cyclotron laboratories followed, the oscillator fed power to the dees by inductive coupling, in the manner of a transformer, and the system was tuned by adjusting the inductance of the coil on the secondary or dee side. Stiff copper rods supported the dees and sat on pyrex tubes carried by the chamber walls. These pyrex tubes were the insulators whose failure gave so much trouble with the 27-inch cyclotron. Lawrence and Cooksey realized that the stopgaps they employed to block stray cathode rays and to cool the glass could not be extended to much higher voltages. When necessary, they supposed, they might adapt Sloan's system of vacuum insulation, in which the secondary coil hangs from a point on its length where the voltage induced in it has a node. But they did not say how they would proceed.^[55] Insulation was not the only

problem in the old dee system. As we know, it transferred only about 50 percent of the power developed by the oscillator into the dee circuit and it returned an unwanted, destabilizing feedback.^[56]

Transmission lines resolved or moderated these problems. Green and Kruger at Illinois improved stability and annihilated feedback by coupling a coaxial line between the oscillator and the dee circuit. The innovation had the additional advantage of distancing the bulky oscillator from the work area around the cyclotron. But it did not bring more power from the oscillator to the dees and did not relieve the stress on the insulators.^[57] The solution, as developed by Dunning and Anderson at Columbia, was to introduce a second set of transmission lines and to make the dees part of them. The scheme may be clear from figure 6.5, in which the coaxial lines coupled to the oscillator as in the Illinois plan are in turn coupled inductively to straight conductors or stems that open out to become the dees. The stems, copper pipes a few inches in diameter, are fixed coaxially in much larger steel cylinders by movable metallic disks. Plate 6.1 shows the cylinders and the dees whose stems they carry, for the MIT cyclotron built by Livingston. Each pipe-and-cylinder pair constitutes a transmission line in which a standing electrical wave may be set up. (The whole may be likened to a transmission line with infinite load: the dees and the stems make up the central wire and the chamber walls and cylinders the outer sheath.) By a clever choice of dimensions and electrical parameters, the cyclotroneer can arrange that the voltage maximum comes at the dees and no voltage exists where the disk or "spider" grips the stem. The cylinder may then be maintained at ground potential. Evidently the distance from the dee tip to the voltage node must be approximately one-quarter of the wavelength at which the cyclotron resonates. Hence the designation of the pipe-and-cylinder as a "l /4 transmission line."^[58]

[Full Size]

Fig. 6.5
Transmission-line coupling of the oscillator to the dee stems. The
nonresonant line from the oscillator taps into the quarter-wave
line ending in the dees. Nahmias,  Machines atomiques , 53.

Dunning and Anderson recommended their double application of transmission-line technology for its greater efficiency, higher voltages, and higher frequencies; its elimination of insulators, shielding of cables, and ease of adjustment; in short, at the bottom line of cyclotroneering, for bringing a "considerable increase in energy and intensity of the ion beam." In the summer of 1938 the Laboratory's directing committee in conclave cyclotron.^[59]

The astounding result—dimensions cyclotroneers could "hardly swallow"—appears in plates 6.2 and 6.3.^[60] In plate 6.2 McMillan sits on the external surface of the left dee support while Alvarez sprawls on the upper pole of the magnet; the large segmented cylinders entering the dee supports are the untuned transmission lines coupling the oscillator to the dee system. In plate 6.3 Green stands with one hand on the mechanism that adjusts the position of the left dee stem while Cooksey looks on from a bower of tubes and cables. The untuned transmission lines now run into a metal housing resting on the external dee supports; their other ends enter a large cabinet on the mezzanine that contains the oscillator. The long black hose penetrating the enlarged cylindrical attachment to the dee system above Green's head provides the power for the beam deflector.

The complexity of the transmission lines and dee supports, the tedious determination by calculation and trial of its dimensions and electrical constants, was perhaps the most difficult challenge for the builders of the Crocker cyclotron. In their success they had the satisfaction of outdoing Harvard, which required Salisbury's services after its radio frequency men had failed to couple their oscillator to their quarter-wave dee supports.^[61] The dee system was but one of many headaches. There were the oscillator itself, Salisbury's larger and stabler version of the 37-inch system, a self-excited oscillator tapped into the dee stems through choke coils at nodes located by cut-and-try;^[62] the vacuum pumps for the oscillator and the chamber, improved by Sloan to a capacity of 5,000 liters/second;^[63] and 10,000 details of wiring, cooling, controlling, and protecting the machine.

A cyclotron is more than the sum of its parts. Although the major components of the 60-inch had proved themselves separately, a beam did not issue from the whole merely by turning on the switches. Beam hunting began at least as early as February 1939, when "most everybody in the laboratory . . . [was] working day and night on the medical cyclotron." All the planning promised an early success, and Lawrence arranged to announce the start up over CBS on April 15. But parasitics plagued operation. Once Salisbury had suppressed them, the transmission system showed itself unable to put sufficient power on the dees. No resonance had occurred by April 4. Cooksey cancelled the broadcast. Alterations in the lengths and positions of the lines improved their behavior and disclosed malfunctions in the deflector system and oscillator circuit. After four months of coaxing there was still no beam.^[64]

Toward the middle of May, a Geiger counter detected a rudimentary proton stream. With much shimming and trimming they made it to the target, a few µA strong, at 7.5 MeV, with only 10 kW on the oscillator. The transmission lines behaved beautifully, delivering 200 kV across the dees with an input of 80 kW. (Ultimately they helped the 60-inch to consume less power than the 37-inch for similar currents: for 70 µA of 8 MeV deuterons the smaller machine needed 45 kW to the magnet and 70 kW to the oscillator; for 80 µA of 16 MeV deuterons, the larger also needed 45 kW to the magnet but only 40 kW on the oscillator.)^[65] Probes showed a milliamp of positive ions within the chamber. Lawrence was much gratified: "We will obtain really prodigious circulating currents," he boasted justifiably to Cockcroft.^[66] In June the first deuterons arrived, at 19 MeV. The reactions of the proprietors were characteristic. Cooksey: "They stick out in the air 153 centimeters. Oh boy!!" Lawrence: "The only difficulty in the way of going on up to a hundred million volts is the financial one." The intent to build bigger surfaced in the announcement of the beam in the Physical Review .^[67]

During July and August the new machine was tuned to higher currents: from 5 µA of 16 MeV of deuterons at the end of June to 15 µA on July 26, with a filament source; and from 50 µA a month later to 100 µA or more in early October, with a capillary. The neutron yield per µA was five times as great at 16 MeV than at the old maximum of 8 MeV.^[68] During the first three weeks of September, a shield of water tanks was installed for protection against neutrons and bombardments began with 15.8 MeV deuterons. The local press was admitted and amazed. The staff, who

had "all decided," according to Cooksey, to bring the crew system to the Crocker, organized to squash all remaining bugs and to build a treatment room before neutron therapy began. Altogether in its twelve weeks' running in, the new cyclotron operated for 155 hours (about half for physicists and half for biologists), stopped for construction for 300 hours (primarily for shielding and building the treatment room), and shut down for repairs and adjustments (mainly to the ion source, electrical controls, and oscillator) for 260 hours.^[69]

Although the medical cyclotron had shown what it could do, it had not contrived to do it steadily by the end of the year. Around Thanksgiving, Kamen had to advise a supplicant for Fe⁵⁹ that "the new machine still continues to be a creature of chance and in its present conditition cannot be expected to run continuously for another month." Nor did it then. It did not work steadily enough to bake probes and could not be depended on for neutron therapy. It suffered from what its operators liked to call "morning sickness," little difficulties with obscure symptoms, like stuck relays and failed heaters, that neophytes could not diagnose. It also had major problems, like shorts in the magnet coils, that kept it in bed for weeks.^[70]

From its resumption of service in mid February, it performed brilliantly, with "almost unbelievably large currents," as much as 200 µA, far more than the targets could take; it ran "with amazing smoothness and stability," seldom missing a scheduled bombardment.^[71] As Lawrence wrote to J.S. Foster, who was trying to secure money to build a cyclotron at McGill, the consequence of this steady, high-energy, high-current bombardment was an output

of neutrons and radioactive materials "no less than prodigious."^[72] The figures in table 6.2 will assist in the estimation of the prodigy.

The intensity of the source increases with the duration of the bombardment until the specimen has been irradiated for a time approximately equal to its period of decay. The figures in the last two columns of table 6.2 usually represent a "day's" bombardment, which might mean anything from eight to twenty-four hours. (Table 6.3 gives more precise yields, in µCi/µAh, for the 60-inch in 1941.) The expense of running the machines was reckoned as follows. The direct cost of operation of the 60-inch averaged out to a little over 10 cents a microamp hour. Now the total capital cost of the 60-inch cyclotron through September 1939 was $63,500; of the labor by then lavished on it, $20,750; and of the Crocker building, $75,000. We have in all about $160,000.^[73] Depreciating this investment over eight years, as the Laboratory did, and reckoning 2,400 hours of running time each year, we arrive at an amortization of $8.33 an hour. The costs—power, supplies, target preparation—for an hour's operation at 100 µA was something over 100 times 10 cents, say $10.50. Maintenance and repairs added another $6 an hour. The real cost of neutron therapy or isotope manufacture at the 60-inch therefore came to about $25.00 an hour, some sixty times the salaries of its attendants.^[74]

2—
Breeding at Home

From Physics to Physic

The multiplication of cyclotrons in the United States began in 1934, with Livingston's little machine at Cornell (plate 6.4). No fewer than eight were commissioned the following year at Bartol,

Columbia, Illinois, Michigan, Princeton, Purdue, Rochester, and Washington. They came in two almost standard sizes: poles around 16 inches in diameter (Cornell, Illinois, Rochester, Washington) and poles around 35 inches in diameter (Bartol, Columbia, Princeton, Purdue). Michigan exceeded them all, and Berkeley too, before the 60-inch, with poles of 42 inches, became the standard for the cyclotrons of the later 1930s (table 6.4). As we know, Lawrence cooperated fully with the builders of these machines, most of whom were his students and associates. Neither he nor the Research Corporation had any interest in restricting the spread of their patented technology.

These first American cyclotrons were made for physics. An indication of their orientation is that Livingston designed the Cornell cyclotron for acceleration of protons and deuterons without giving priority to either; he had separate chambers available for each from the time his cyclotron began to operate in 1935. Although deuterons, the particles of preference at Berkeley, are particularly effective in making radioisotopes and generating neutron beams,^[75] protons are better for close examination and

interpretation of nuclear reactions and forces. Moreover deuterons, which have a lower resonant frequency in the cyclotron than protons, can be accelerated to relatively higher energies with relatively less bother about the always painful oscillating circuit. As medical and biological applications of cyclotrons became paramount in acquiring financial support for their construction, parameters were set to favor the production of deuterons. Most of the large cyclotrons in operation in 1940 were not suitable for the acceleration of protons. In this circumstance Hans Bethe saw cause for regret and even alarm.^[76] But with his colleague Living-

ston's cyclotron, which delivered 3 or 4 µA of 2.0 MeV protons (or 1.35 MeV deuterons), Bethe could test the predictions of yields of nuclear reactions in light elements computed by himself and his students on the basis of Bohr's theory of nuclear structure.

The only applications envisaged by Livingston in his first description of the Cornell cyclotron concerned nuclear problems, viz., the study of "all types of reactions" caused by "all available bombarding particles." At first Cornell used deuterons in the reaction Be(d,a )Li in order to infer the states of the beryllium nucleus from the energies of its ejected alpha particles; they subsequently exploited the reverse transformation Li(d,n)Be for neutrons for scattering experiments; and they ended, in the late 1930s, with careful measurements of deuteron-induced disintegrations of nitrogen, carbon, and oxygen.^[77] Their allegiance to the original purpose of their machine was secured by its size: it could not make radioisotopes in sufficient quantities for biomedical studies and foundation support.^[78] The same things are true—enforced devotion to physics—of the Illinois cyclotron. Like its twin at Cornell, it was used to generate secondary beams that themselves became the instrument or object of study, for example, gamma rays and neutrons.^[79] That was not enough for cyclotroneers, however, and, as we know, both Green and Kruger at Illinois and Livingston at Cornell made important improvements in the equipment of cyclotrons.^[80]

The first of the larger cyclotrons to be completed outside Berkeley was Princeton's. It was more powerful (several µA of 10 MeV deuterons) and more costly ($12,000 inclusive of motor generator) than Cornell's. It was adaptable to isotope production. Henderson and White emphasized magnet design and explicitly preferred deuterons; their machine, although financed by the university and

not by outside medical interests and dedicated, according to their first description of it, to nuclear physics, proved a useful manufacturer of radioelements, to which it came to devote much of its time. Like Cornell's cyclotron, Princeton's began to run within a year or so of its commissioning.^[81]

Rochester's cyclotron also took but a year to build. Its promoter, Lee DuBridge, "one of the earliest to see a great future for the cyclotron," had been a National Research Fellow at Berkeley, though not in the Radiation Laboratory. He built with medical support and the help of industry and the Research Corporation.^[82] His 20-inch machine was dedicated at first to proton work in the unexplored region above the reach of high-tension apparatus and the Cornell and Illinois cyclotrons. Working at 2 or 3 µA, he and his colleagues reaped a good harvest of the known reactions—p capture and (p,a ) disintegrations—in the "fast" proton range of 1–4 MeV, and they discovered a new process, (p,n), which sets in at about 3 MeV. With this success, they decided to enlarge their magnet's poles to 26 inches in the expectation of having protons of over 7 MeV. They reached 7.2 MeV and followed (p,n) reactions in at least thirty-five elements. As their experimental work continued Cornell's, so their theorist, Victor Weisskopf, continued Bethe's. As at Cornell, theory went hand in hand with experiment.^[83]

The enlargement of the poles, which occurred in 1937, made possible the production of 4.5 MeV deuterons. They sufficed to make radioisotopes in sufficient quantities for tracer work by physicians and biologists at Rochester, and for an investment of

$30,000 by the Rockefeller Foundation, to be split about equally between the physicists and the doctors. The physicists used their share to attract Van Voorhis, who later rearranged the radio frequency system to double the power reaching the dees. The proton bombardments and the physics program retained the ascendency at Rochester, and only one day a month was devoted to deuteron irradiation of materials for biomedical experiments.^[84]

The balance fell out quite differently for cyclotrons of greater power than Rochester's. Columbia and Michigan mark the transition. Columbia began in February 1935, asking Lawrence's opinion about the conversion of a general purpose 14-inch magnet. He advised a magnet and machine of the type—and hence for the purposes—that Livingston was then pursuing at Cornell. An opportunity then presented itself to build bigger, in the form of the navy's 500-kW Poulsen arc, not Berkeley's twin, but big enough to support pole pieces 36 inches in diameter, which NYU decided it could not use. That was in the summer of 1935.^[85] By the time the magnet with its 36-inch poles was in place at the end of 1936, the biomedical economics of cyclotrons had been discovered; Columbia University, in consideration of medical applications, put up money for the completion of the machine and the chairman of the physics department and a colleague from chemistry went to the Research Corporation for more, "to produce very usable quantities of artificial radioactive substances." They got what they wanted, some $1,850.^[86] By the winter of 1938/39, having shimmed away the bad asymmetry of the poles and perfected the quarter-wave transmission line, the Columbia group, which by then included Paxton, was ready to make P³² and Na²⁴ for biological tracing. Eventually they had a neutron flux equivalent to that from 50 kg of Rn-Be and could detect on the

roof of their fourteen-story physics building neutrons that had passed through their colleagues on the way up from the cyclotron in the basement. "The cyclotron really operates very beautifully now."^[87] Between the decision to make a cyclotron, taken in February 1935, and its completion in 1938, the purpose the machine was to serve, the basis of its financing, and the magnitude of its hazards had been transformed.

The Michigan cyclotron was very well supported from the beginning—from the summer of 1935—by the university's Rackham funds. The chairman of the physics department, H.M. Randall, and the machine's main builder, James Cork, who had spent 1935/36 on sabbatical in Berkeley, apparently had indicated to the university's medical school, whose collaboration they obtained, that the large cyclotron they planned would yield radioisotopes for clinical use. That was more than Berkeley had by then achieved, and it was doubtless with relief that Randall learned from Lawrence in the summer of 1936 that the 27-inch could make samples of Na²⁴ equivalent to 50 mg of radium. "You can proceed with the construction of a cyclotron with the definite assurance that you will be able to produce enough radiosodium and other radioactive substances for medical investigations."^[88] The Michigan cyclotron put out its first beam (several µA of 5 MeV deuterons) in August 1936, about a year after the physics department had assigned it first priority. Lawrence, who happened to be in town at start up, "got a great [and well-deserved] kick out of seeing the second large cyclotron in the world go into operation." A new tank installed with Thornton's help in the winter of 1937/38 gave an intense beam that reached to 200 µA in the summer of 1939. That more than met the debt to the medical people, who had been satisfied with 15 µA.^[89]

Apart from the 27-inch cyclotrons begun at Yale in 1937 and at Stanford in 1940, built on the cheap for experiments in physics,^[90] all the machines commissioned from 1936 on were at least as large as Michigan's and, like it, dedicated largely to biomedical work. Physics became a secondary or tertiary goal of cyclotron builders, according to their reviews of their discipline. Kurie, Chadwick, and Mann, for example, gave pride of place to radiobiology, chemical tracing, and clinical treatments in their accounts of uses of the cyclotron; and Mann, writing in Nature in April 1939, offered Kamen and Wilson's internal target, which multiplied the harvest of therapeutically relevant isotopes, as "the most important recent" improvement in the cyclotron art and a grand "stimulus to those who are in favour of the immediate construction of a cyclotron for medical purposes in Great Britain."^[91] Lawrence was also a stimulus in this cause. His recommendation to the British that their medical cyclotron have 50-inch poles ("[the 60-inch] is probably larger than necessary"), might suggest that machines of the Michigan class were not big enough to practice medicine. It appears that 50 inches was an arithmetical compromise between the excessive 60-inch and the adequate Michigan type: as Lawrence wrote Livingood, "a 40-inch cyclotron really fills the bill for most medical purposes." Or, to put the point backWard, since the 37-inch was adequate for therapy and isotope manufacture, a 50-inch would do.^[92]

Lawrence's optimum medical cyclotron, although costly, came to considerably less than the Crocker machine. Late in 1938 someone at the Laboratory took the trouble to estimate the differences in price: for the magnet, metal and installation, $40,000 for the 60-inch, $18,000 for the 50-inch; for the remaining parts, $40,000 versus $32,000; for labor, $8,000 and $6,000. Since something could be saved by making some parts in departmental shops and by gaining concessions from manufacturers, Lawrence quoted $50,000 in all as the ante "to do it right from the standpoint of clinical work." But if only $20,000 were available, it might just do.^[93]

The largest cyclotron started in the United States after the Crocker was almost its identical twin, erected by, of all people, Lawrence's again friendly rival Merle Tuve. The Carnegie Institution of Washington, with its high investment in high tension, could scarcely remain aloof from the new forces and opportunities in high-energy physics. "If you want an equipment for producing artificial activity [as opposed to one for exact physics], there is nothing that even faintly begins to compare with a cyclotron, nor has there ever been." So Tuve answered an enquiry from Bell Labs in the spring of 1940. By then he had been cyclotroneering for a year. During the spring of 1939, Brobeck, Cooksey, and John Lawrence joined interested parties in the Washington area to discuss the desirability of building a big cyclotron at Carnegie for service to local institutions. The consensus held that Carnegie could do no better than to build a 60-inch machine. As the director of Tuve's unit explained to the president of the Carnegie Institution, the machine could do physics for the Department of Terrestrial Magnetism and chemistry and biology for other units of the institution, local universities (Johns Hopkins, George Washington, Catholic) and the federal government (the Department of Agriculture, the National Cancer Institute).^[94] It would not practice medicine. "Medical research will be given no preference whatsoever and, in fact, will be undertaken only in the sense of

fundamental work in physiology and biochemistry." Physics would have a third of the machine time.^[95] The Carnegie philanthropies had their own resources and freedom to determine their own programs.

Once commissioned, the Carnegie cyclotron came into existence in the manner then standard. Tuve and his main associate, Richard Roberts, received an "enormous roll of blue prints" from Berkeley, "almost lost our mind[s]" reading through them, learned that a visit to Berkeley and Berkeley cyclotroneers was indispensable, hired two (Green and Abelson), and started shopping for steel.^[96] They differed from Berkeley in contracting with ARMCO for the magnet ($16,025), part of which was cast (as at Chicago, Columbia (pole tips only), Harvard, MIT, and Ohio State, among others), and with GE for water-cooled coils to run it (as at Columbia, Harvard, MIT, and Rochester). Nonetheless, the whole would resemble the Crocker magnet very closely, in weight as well as in shape, the coils like "the turret[s] of a battle ship."^[97] The analogy was not idle. Like all other cyclotron builders in 1939 and 1940, the Carnegie Institution faced the free-market competition of an arms race. "War orders have made loads of extra work in getting contracts that are within starting distances of estimates based on Berkeley costs."^[98]

Cyclorama

There were twenty-two cyclotrons completed or under construction in the United States in 1940. They came in four sizes, as indicated in table 6.5, which also contains information about their coming-to-be. As the table shows, the quantity of steel and copper in the magnet increases much more rapidly than the diameter of its poles, somewhere between the cube and the square. Since the

magnet's metal was the single largest expense in the construction of a cyclotron, its unit cost became a matter of concern as prices rose owing to inflation in the worldwide buildup of arms. In 1935 mild steel boiler plate, which Cook and Henderson regarded as almost the equivalent of ARMCO iron, cost 1.6 cents a pound. With iron so cheap, Lawrence wrote, a good magnet could be made from scratch more cheaply than refurbishing an old navy arc, and he urged DuBridge to build bigger—the Rochester cyclotron initially was to have 14-inch poles—since the raw ingredients came for a song.^[99] In 1937 the prices of steel and copper were about twice what they had been in 1935, and in 1938 they increased another 150 percent. The implicit and symbolic competition between cyclotrons and armaments for strategic materials in the late 1930s became explicit and realistic in the early 1940s, when several incomplete machines secured high-priority allocations of iron and copper in the national interest.^[100] Only medicine could cope with the rapidly inflating capital requirements of cyclotroneers.

The second set of information in table 6.5 reveals that, with a few exceptions easily explained away, a baby cyclotron took about a year to build, a small cyclotron perhaps a year and a half, a medium one two years or a little more. The exceptions: Washington, a three-year birth in the baby class, was almost entirely the work of one man; Yale, two years acoming in the small class, also had a very small crew and assembled its own magnet; in the middle class, Columbia was delayed by navy bureaucracy and by its own innovativeness, while Purdue suffered from lack of resources, and both had the disadvantage of doing almost entirely without a man from Berkeley.^[101] But a physicist or two who knew what they were about, did not hanker after novelties, and had the help of a graduate student, a competent shop, and enough money for their project, could bring a 90-ton cyclotron from the drawing boards to first beam in two years or less. The record of Berkeley men in partibus may be read from table 6.5.

How many cyclotrons did the United States require? Of what sizes and capacities? In May 1938 Lawrence had an easy recipe: "There should be a cyclotron laboratory in every university center which will provide ammunition for unending work in nuclear physics and biology as well as in clinical medicine." Cooksey explained the situation to Karl Darrow, an industrial physicist and physics popularizer, who calculated that "the country may need a thousand cyclotrons." In October 1940 Urey judged that the country had as many as it needed, or could afford. "There are cyclotrons and Van de Graaff machines in most of the universities of the United States. . . . Some institutes have one or two of each."^[102] The natural limit to their reproduction might well have been sighted. Their costs were increasing as the square or cube of their size and their number perhaps linearly in time, while the national funding base showed no prospect of rapid enlargement. Indeed, there were signs of retrenchment: the National Advisory Cancer Council declined to distribute $100,000 for the capital improvement of cyclotrons, as Lawrence counselled in 1938 and 1939; the Research Corporation, with a reduced income, cut off support to leading cyclotron laboratories like Rochester in 1939 and 1940; the Rockefeller Foundation turned down Harvard and looked askance at all applications in support of new cyclotrons that did not go beyond the reach of Berkeley's 60-inch. The sovereign remedy for weak finances—"getting the phosphorus up in millicuries will bring you the [needed] backing, and support"—had, by repetition, made foundations resistant.^[103]

In this Malthusian situation, the American Institute of Physics and the cyclotroneers at Harvard and MIT called a conference on applied nuclear physics that met in Cambridge from October 28 to November 2, 1940. For it Livingston drew up the chart reproduced as table 6.6. Apart from the baby cyclotrons, installation costs, including labor, did increase with at least the square of the pole size, whereas the operating expense of small and medium

cyclotrons came to about the same.^[104] The last represented a true gain in efficiency: the typical medium machine needed less tending than the smaller ones and consumed no more power. For a long time the price of power was the most worrisome part of

Lawrence's budget: the 27-inch ate up around $1,500 a year in 1933/34 and 1934/35 and almost twice that in 1935/36; after its enlargement to 37 inches, it required as much as 50 kW for the magnet, and (for 100 µA of deuterons) around 40 kW for the oscillator, which, at the Laboratory's cost of 2 cents/kWh and at Livingston's figure of 2,400 operating hours a year, amounted to almost $4,500 per annum.^[105] The Harvard and MIT cyclotrons—made identical in size to "short circuit [covetousness]"—required only $1,500 a year for power, about a third that of the 37-inch. Their more compact magnets of ARMCO iron, transmission lines, and lower unit costs (around 1.5 cents/kWh) compassed the reduction.^[106]

The most significant figures in Livingston's table concern output. From the beam currents (numbers to the right of the slash are Livingston's estimates of the expected eventual performance of the Harvard and MIT machines), the assumed operation of seven hours a day, and the operating costs, the unit price of the common tracer P³² is readily deduced. The advantage of the medium machines leaps to the eye: when in stride, the Cambridge 42-inchers would produce P³² at about $1/mCi, very much cheaper than the Rochester cyclotron could do and not much more than the cost at the Crocker. (These figures must be taken as approximations; from operational data on the Berkeley machines, the Carnegie Institution deduced that a mCi would cost $6.50 at the 37-inch cyclotron and $2.25 at the 60-inch.)^[107] A similar story emerges from Livingston's figures on neutron production. From the neutron intensity in equivalents of Rn-Be follows the neutron effect in "n units" per minute at one meter from the cyclotron's

beryllium target; and from the n units and the hypothesis that energy delivered by neutrons to living tissue is four times as destructive as the same quantity of energy delivered by x or gamma rays, Livingston arrived at the penultimate line of his chart. The bottom line, the cost of therapy per 100 roentgens/minute, shows that price did not limit the effective treatment of cancer by neutrons.^[108]

To bring out clearly the relative excellence of the cyclotron as a factory for radioisotopes and therapeutic neutrons, Livingston rated the electrostatic generators of 1940 as shown in table 6.7. A comparison of the three classes of generators with the largest classes of cyclotrons (table 6.6) shows that although the cyclotrons cost more than the corresponding Van de Graaffs in both capital investment and operating expenses, they enjoyed so great an advantage in beam and energy that they manufactured radioisotopes and neutron doses at much lower unit prices. The standard moderate cyclotron made—or could make, after improvement—a millicurie of radiophosphorus in six minutes for less than a dollar; the top-of-the-line generator, Tuve's 19-foot pressurized Van de Graaff, also after perfection, would need about twenty hours—and $166—to do the same. It was not that Van de Graaff and his associates had been idle or ill-financed. Between 1936 and 1940 both the cyclotron and the generator increased their effective beam energies by a factor of four and beam currents by a factor of ten or more: from 4 to 16 MeV and 20 to 200 µA at Berkeley, from 0.9 to 3.5 MeV at the Carnegie Institution, and from the Carnegie's 10 µA to MIT's 100 µA and more.^[109] But an electrostatic generator that could hold 3.5 MV was a technological freak—Westinghouse was trying for 5 MV but could scarcely reach 3—and MIT's Van de Graaff represented the effective energy limit in 1940 if a respectable current were desired.^[110] Clinical doses of radioisotopes lay beyond its capabilities.

It is not easy to state the goals of the leaders of cyclotron laboratories in 1940. On the one hand, their instruments had opened up vast fields of biological and medical research, held promise of discoveries in nuclear chemistry and physics, and constantly challenged and enticed Homo faber . Theirs was an exciting and progressive line of work. On the other hand, the cyclotron brought slavery to physicians and to the chase for money, regimentation of laboratory work, and no long-range research project.

It was a tool constantly in need of improvement lest it condemn itself and its attendants to routine manufacture in the service of others. After consulting with Lawrence and Conant, Karl Compton summed up the situation: "To maintain an active program and a well rounded staff has required more aggressive salesmanship than the scientific profession relishes. . . . , an abnormal competitive element which is unfortunate."^[111] The cyclotroneers escaped the logic of their situation—an increasingly competitive struggle for large sums in an increasingly inelastic market, a growing disparity between builder-physicists and operator-technicians, a tightening tension between service to others and science for oneself—by going off to war.