Atomic Energy for Military Purposes (The Smyth Report)



4.1. In Chapter II the general problems involved in producing a chain reaction for military purposes were described. Early in the summer of 1940 the questions of most immediate importance were:

(1) Could any circumstances be found under which the chain reaction would go?

(2) Could the isotope U-235 be separated on a large scale?

(3) Could moderator and other materials be obtained in sufficient purity and quantity?

Although there were many subsidiary problems, as will appear in the account of the progress made in the succeeding eighteen months, these three questions determined the course of the work.



4.2. In June 1940, nearly all work on the chain reaction was concentrated at Columbia under the general leadership of Pegram, with Fermi and Szilard in immediate charge. It had been concluded that the most easily produced chain reaction was probably that depending on thermal neutron fission in a heterogeneous mixture of graphite and uranium. In the spring of 1940 Fermi, Szilard and H. L. Anderson had improved the accuracy of measurements of the capture cross section of carbon for neutrons, of the resonance (intermediate-speed) absorption of neutrons by U-238, and of the slowing down of neutrons in carbon.

4.3. Pegram, in a memorandum to Briggs on August 14, 1940, wrote, "It is not very easy to measure these quantities with accuracy without the use of large quantities of material. The net results of these experiments in the spring of 1940 were that the possibility of the chain reaction was not definitely proven, while it was still further from being definitely disproven. On the whole, the indications were more favorable than any conclusions that could fairly have been claimed from previous results."

4.4. At a meeting on June 15 (see Chapter III) these results were discussed and it was recommended that (A) further measurements be made on nuclear constants, and (B) experiments be made on lattices of uranium and carbon containing amounts of uranium from one fifth to one quarter the estimated critical amounts.


4.5. Pegram's report of February 15, 1941 shows that most of the work done up to that time was on (A), while (B), the so-called intermediate experiment, was delayed by lack of materials.

4.6. Paraphrasing Pegram's report, the main progress was as follows:

(a) The slowing down of neutrons in graphite was investigated by studying the intensity of activation of various detectors (rhodium, indium, iodine) placed at various positions inside a rectangular graphite column of dimensions 3 X 3 X 8 feet when a source of neutrons was placed therein. By suitable choice of cadmium screens the effects of resonance and thermal neutrons were investigated separately. A mathematical analysis, based on diffusion theory, of the experimental data made it possible to predict the results to be expected in various other arrangements. These results, coupled with theoretical studies of the diffusion of thermal neutrons, laid a basis for future calculations of the number of thermal and resonance neutrons to be found at any point in a graphite mass of given shape when a given neutron source is placed at a specified position within or near the graphite.

(b) The number of neutrons emitted in fission. The experiments on slowing down neutrons showed that high-energy (high-speed) neutrons such as those from fission were practically all reduced to thermal energies (low speeds) after passing through 40 cm or more of graphite. A piece of uranium placed in a region where thermal neutrons are present absorbs the thermal neutrons and -fission occurs- re-emits fast neutrons, which are easily distinguished from the thermal neutrons. By a series of measurements with and without uranium present and with various detectors and absorbers, it is possible to get a value for the constant η, the number of neutrons emitted per thermal neutron absorbed by uranium. This is not the number of neutrons emitted per fission, but is somewhat smaller than that number since not every absorption causes fission.

(c) Lattice theory. Extensive calculations were made on the probable number of neutrons escaping from lattices of various designs and sizes. This was fundamental for the so-called intermediate experiment, mentioned above as item (B).

The presence of neutrons can be detected by ionization chambers or counters or by the artificial radioactivity induced in various metal foils. (See Appendix 1.) The response of each of these detectors depends on the particular characteristics of the detector and on the speed of the neutrons (e.g., neutrons of about 1.5 volts energy are particularly effective in activating indium). Furthermore, certain materials have very large absorption cross sections for neutrons of particular ranges of speed (e.g., cadmium for thermal neutrons). Thus measurements with different detectors with or without various absorbers give some indication of both the number of neutrons present and their energy distribution. However, the state of the art of such measurements is rather crude.


4.7. Early in 1941 interest in the general chain-reaction problem by individuals at Princeton, Chicago, and California led to the approval of certain projects at those institutions. Thereafter the work of these groups was coordinated with the work at Columbia, forming parts of a single large program.


4.8. In Chapter II it is stated that there were advantages in a lattice structure or "pile" with uranium concentrated in lumps regularly distributed in a matrix of moderator. This was the system on which the Columbia group was working. As is so often the case, the fundamental idea is a simple one. If the uranium and the moderator are mixed homogeneously, the neutrons on the average will lose energy in small steps between passages through the uranium so that in the course of their reduction to thermal velocity the chance of their passing through uranium at any given velocity, e.g., at a velocity corresponding to resonance absorption, is great. But, if the uranium is in large lumps spaced at large intervals in the moderator, the amounts of energy lost by neutrons between passages from one lump of uranium to another will be large and the chance of their reaching a uranium lump with energy just equal to the energy of resonance absorption is relatively small. Thus the chance of absorption by U-238 to produce U-239, compared to the chance of absorption as thermal neutrons to cause fission, may be reduced sufficiently to allow a chain reaction to take place. If one knew the exact values of the cross sections of each uranium isotope for each type of absorption and every range of neutron speed, and had similar knowledge for the moderator, one could calculate the "optimum lattice," i.e., the best size, shape and spacing for the lumps of uranium in the matrix of moderator. Since such data were only partially known, a direct experimental approach appeared to be in order. Consequently it was proposed that the absorption of neutrons by uranium should be measured under conditions similar to those expected in a chain-reacting pile employing graphite as moderator.

The term "resonance absorption" is used to describe the very strong absorption of neutrons by U-238 when the neutron energies are in certain definite portions of the energy region from 0 to 1,000 electron volts. Such resonance absorption demonstrates the existence of nuclear energy levels at corresponding energies. On some occasions the term resonance absorption is used to refer to the whole energy region in the neighborhood of such levels.

4.9. Experiments of this type were initiated at Columbia, and were continued at Princeton in February 1941. Essentially the experiment consisted of studying the absorption of neutrons in the energy range extending from a few thousand electron volts down to a fraction of an electron volt (thermal energies), the absorption taking place in different layers of uranium or uranium oxide spheres embedded in a pile of graphite.

4.10. In these experiments, a source of neutrons was provided by a mean of protons (accelerated by a cyclotron) impinging on a beryllium target. (The resulting yield of neutrons was equivalent to the yield from a radium-beryllium source of about 3,500 curies strength.) The neutrons thus produced had a wide, continuous, velocity distribution. They proceeded from this source into a large block of graphite. By placing the various uranium or uranium-oxide spheres inside the graphite block at various positions representing increasing distances from the source, absorption of neutrons of decreasing average speeds down to thermal speeds was studied. It was found that the total absorption of neutrons by such spheres could be expressed in terms of a "surface" effect and a "mass" effect.

4.11. These experiments, involving a variety of sphere sizes, densities, and positions were continued until the spring of 1942, when most of the group was moved to Chicago. Similar experiments performed at a later date at the University of Indiana by A. C. G. Mitchell and his co-workers have verified and in some cases corrected the Princeton data, but the Princeton data were sufficiently accurate by the summer of 1941 to be used in planning the intermediate-pile experiments and the subsequent experiments on operating piles.

4.12. The experimental work on resonance absorption at Princeton was done by R. R. Wilson, E. C. Creutz, and their collaborators, under the general leadership of H. D. Smyth; they benefited from the constant help of Wigner and Wheeler and frequent conferences with the Columbia group.


4.13. About July 1941 the first lattice structure of graphite and uranium was set up at Columbia. It was a graphite cube about 8 feet on an edge, and contained about 7 tons of uranium oxide in iron containers distributed at equal intervals throughout the graphite. A preliminary set of measurements was made on this structure in August 1941. Similar structures of somewhat larger size were set up and investigated during September and October, and the so-called exponential method (described below) of determining the multiplication factor was developed and first applied. This work was done by Fermi and his assistants, H. L. Anderson, B. Feld, G. Weil, and W. H. Zinno

4.14. The multiplication-factor experiment is rather similar to that already outlined for the determination of η, the number of neutrons produced per thermal neutron absorbed. A radium-beryllium neutron source is placed near the bottom of the lattice structure and the number of neutrons is measured at various points throughout the lattice. These numbers are then compared with the corresponding numbers determined when no uranium is present in the graphite mass. Evidently the absorption of neutrons by U-238 to produce U-239 tends to reduce the number of neutrons, while the fissions tend to increase the number. The question is: Which predominates? or, more precisely, Does the fission production of neutrons predominate over all neutron-removal processes other than escape? Interpretation of the experimental data on this crucial question involves many corrections, calculations, and approximations, but all reduce in the end to a single number, the multiplication factor k.


4.15. The whole success or failure of the uranium project depended on the multiplication factor k, sometimes called the reproduction factor. If k could be made greater than 1 in a practical system, the project would succeed; if not, the chain reaction would never be more than a dream. This is clear from the following discussion, which applies to any system containing fissionable material. Suppose that there is a certain number of free neutrons present in the system at a given time. Some of these neutrons will themselves initiate fissions and will thus directly produce new neutrons. The multiplication factor k is the ratio of the number of these new neutrons to the number of free neutrons originally present. Thus, if in a given pile comprising uranium, carbon, impurities, containers, etc., 100 neutrons are produced by fission, some will escape, some will be absorbed in the uranium without causing fission, some will be absorbed in the carbon, in the containers or in impurities, and some will cause fission, thereby producing more neutrons (See drawing). If the fissions are sufficiently numerous and sufficiently effective individually, more than 100 new neutrons will be produced and the system is chain reacting. If the number of new neutrons is 105, k = 1.05. But if the number of new neutrons per 100 initial ones is 99, k = .99 and no chain reaction can maintain itself.

4.16. Recognizing that the intermediate or "exponential" experiment described above was too small to be chain reacting, we see that it was a matter of great interest whether any larger pile of the same lattice structure would be chain reacting. This could be determined by calculating what the value of k would be for an infinitely large lattice of this same type. In other words, the problem was to calculate what the value of k would be if no neutrons leaked away through the sides of the pile. Actually it is found that, once a chain-reacting system is well above the critical size - say two or three times as great - and is surrounded by what is called a reflector, the effective value of k differs very little from that for infinite size provided that k is near 1.00. Consequently, it has become customary to characterize the chain-reaction potentialities of different mixtures of metal and moderator by the value of k the multiplication constant obtained by assuming infinite size of pile.

4.17. The value of k as reported by Fermi to the Uranium Section in the fall of 1941 was about 0.87. This was based on results from the second Columbia intermediate experiment. All agreed that the multiplication factor could be increased by greater purity of materials, different lattice arrangements, etc. None could say with certainty that it could be made greater than one.


4.18. At about the same time that the work on resonance; absorption was started at Princeton, S. K. Allison, at the suggestion of A. H. Compton, began work at Chicago under a contract running from January 1, 1941 to August 1, 1941. The stated objectives of the work were to investigate (a) the increase in neutron production when the pile is enclosed in a beryllium envelope or "reflector," and (b) the cross sections of beryllium. A new contract was authorized on July 18, 1941, to run to June 30, 1942. This stated the somewhat broader objective of investigating uranium-beryllium-carbon systems generally. The appropriations involved were modest: $9,500 for the first contract, and $30,000 for the second contract.

4.19. As has already been pointed out in Chapter II, beryllium has desirable qualities as a moderator because of its low atomic weight and low neutron-absorption cross section; there was also the possibility that a contribution to the number of neutrons would be realized from the (n, 2n) reaction in beryllium. The value of the cross section was not precisely known; furthermore it was far from certain that any large amount of pure beryllium could be obtained. Allison's problem was essentially similar to the Columbia problem, except for the use of beryllium in place of graphite. Because of the scarcity of beryllium it was suggested that it might be used in conjunction with graphite or some other moderator, possibly as a reflector.

4.20. In the Chicago experiments, neutrons produced with the aid of a cyclotron were caused to enter a pile of graphite and beryllium. Allison made a number of measurements on the slowing down and absorption by graphite which were valuable checks on similar experiments at Columbia. He finally was able to obtain enough beryllium to make significant measurements which showed that beryllium was a possible moderator comparable to graphite. However, beryllium was not in fact used at all extensively in view of the great difficulty of producing it in quantity in the required structural forms.

4.21. This Chicago project as described above became part of the Metallurgical Laboratory project established at the University of Chicago early in 1942.


4.22. Both the intermediate experiments at Columbia and the continued resonance-absorption work at Princeton required skillful theoretical interpretation. Fermi worked out the theory of the "exponential" pile and Wigner the theory of resonance absorption; both these men were constantly conferring and contributing to many problems. Wheeler of Princeton, Breit of Wisconsin, and Eckart of Chicago - to mention only a few-also made contributions to general pile theory and related topics. Altogether one can say that by the end of 1941 the general theory of the chain reaction for slow neutrons was almost completely understood. It was the numerical constants and technological possibilities that were still uncertain.

4.23. On the theory of a fast-neutron reaction in U-235 a good deal of progress had also been made. In particular, new estimates of the critical size were made, and it was predicted that possibly 10 per cent of the total energy might be released explosively. On this basis one kilogram of U-235 would be equivalent to 2,000 tons of TNT. The conclusions are reviewed below in connection with the National Academy Report. It is to be remembered that there are two factors involved: (1) how large a fraction of the available fission energy will be released before the reaction stops; (2) how destructive such a highly concentrated explosion will be.


4.24. In Chapter I mention is made of the suggestion that the element 94, later christened plutonium, would be formed by beta-ray disintegrations of U-239 resulting from neutron absorption by U-238 and that plutonium would probably be an alpha-particle emitter of long half-life and would undergo fission when bombarded by neutrons. In the summer of 1940 the nuclear physics group at the University of California in Berkeley was urged to use neutrons from its powerful cyclotron for the production of plutonium, and to separate it from uranium and investigate its fission properties. Various pertinent experiments were performed by E. Segré, G. T. Seaborg, J. W. Kennedy, and A. C. Wahl at Berkeley prior to 1941 and were reported by E. O. Lawrence to the National Academy Committee (see below) in May 1941 and also in a memorandum that was incorporated in the Committee's second report dated July 11,1941. It will be seen that this memorandum includes one important idea not specifically emphasized by others (paragraph 1.58), namely, the production of large quantities of plutonium for use in a bomb.

4.25. We quote from Lawrence's memorandum as follows: "Since the first report of the National Academy of Sciences Committee on Atomic Fission, an extremely important new possibility has been opened for the exploitation of the chain reaction with unseparated isotopes of uranium. Experiments in the Radiation Laboratory of the University of California have indicated (a) that element 94 is formed as a result of capture of a neutron by uranium 238 followed by two successive beta-transformations, and furthermore (b) that this transuranic element undergoes slow neutron fission and therefore presumably behaves like uranium 235.

"It appears accordingly that, if a chain reaction with unseparated isotopes is achieved, it may be allowed to proceed violently for a period of time for the express purpose of manufacturing element 94 in substantial amounts. This material could be extracted by ordinary chemistry and would presumably be the equivalent of uranium 235 for chain reaction purposes.

"If this is so, the following three outstanding important possibilities are opened:

"1. Uranium 238 would be available for energy production, thus increasing about one hundred fold the total atomic energy obtainable from a given quantity of uranium.

"2. Using element 94 one may envisage preparation of small chain reaction units for power purposes weighing perhaps a hundred pounds instead of a hundred tons as probably would be necessary for units using natural uranium.

"3. If large amounts of element 94 were available it is likely that a chain reaction with fast neutrons could be produced. In such a reaction the energy would be released at an explosive rate which might be described as 'super bomb.' "


4.26. As previously stated, the fragments resulting from fission are in most cases unstable nuclei, that is, artificially radioactive materials. It is common knowledge that the radiations from radioactive materials have deadly effects akin to the effects of X-rays.

4.27. In a chain-reacting pile these radioactive fission products build up as the reaction proceeds. (They have, in practice, turned out to be the most troublesome feature of a reacting pile.) Since they differ chemically from the uranium, it should be possible to extract them and use them like a particularly vicious form of poison gas. This idea was mentioned in the National Academy report (see paragraph 4.48) and was developed in a report written December 10, 1941, by E. Wigner and H. D. Smyth, who concluded that the fission products produced in one day's run of a 100,000 kw chain-reacting pile might be sufficient to make a large area uninhabitable.

4.28. Wigner and Smyth did not recommend the use of radioactive poisons nor has such use been seriously proposed since by the responsible authorities, but serious consideration was given to the possibility that the Germans might make surprise use of radioactive poisons, and accordingly defensive measures were planned.



4.29. In Chapter I the attribution of thermal-neutron fission of uranium to the U-235 isotope was mentioned as being experimentally established. This was done by partly separating minute quantities of the uranium isotopes in A. O. Nier's mass spectrograph and then studying the nuclear properties of the samples. Additional small samples were furnished by Nier in the summer of 1941 and studied by N. P. Heydenburg and others at M. A. Tuve's laboratory at the Department of Terrestrial Magnetism of the Carnegie Institution of Washington. But results of such experiments were still preliminary, and it was evident that further study of larger and more completely separated samples was desirable.

4.30. The need of larger samples of U-235 stimulated E. O. Lawrence at Berkeley to work on electromagnetic separation. He was remarkably successful and by December 6, 1941 reported that he could deposit in one hour one microgram of U-235 from which a large proportion of the U-238 had been removed.

4.31. Previously, at a meeting of the Uranium Committee, Smyth of Princeton had raised the question of possible large-scale separation of isotopes by electromagnetic means but had been told that it had been investigated and was considered impossible. Nevertheless, Smyth and Lawrence at a chance meeting in October 1941 discussed the problem and agreed that it might yet be possible. Smyth again raised the question at a meeting of the Uranium Committee on December 6 and at the next meeting (December 18, 1941) there was a general discussion of large-scale electromagnetic methods in connection with Lawrence's report of his results already mentioned. The consequences of this discussion are reported in Chapter XI.


4.32. Though we have made it clear that the separation of U-235 from U-238 might be fundamental to the whole success of the project, little has been said about work in this field. Such work had been going on since the summer of 1940 under the general direction of H. C. Urey at Columbia. Since this part of the uranium work was not very much affected by the reorganization in December 1941, a detailed account of the work is reserved for Chapters IX and X. Only a summary is presented here.

4.33. After careful review and a considerable amount of experimenting on other methods, it had been concluded that the two most promising methods of separating large quantities of U-235 from U-238 were by the use of centrifuges and by the use of diffusion through porous barriers. In the centrifuge, the forces acting on the two isotopes are slightly different because of their differences in mass. In the diffusion through barriers, the rates of diffusion are slightly different for the two isotopes, again because of their differences in mass. Each method required the uranium to be in gaseous form, which was an immediate and serious limitation since the only suitable gaseous compound of uranium then known was uranium hexafluoride. In each method the amount of enrichment to be expected in a single production unit or "stage" was very small; this indicated that many successive stages would be necessary if a high degree of enrichment was to be attained.

4.34. By the end of 1941 each method had been experimentally demonstrated in principle; that is, single-stage separators had effected the enrichment of the U-235 on a laboratory scale to about the degree predicted theoretically. K. Cohen of Columbia and others had developed the theory for the single units and for the series or "cascade" of units that would be needed. Thus it was possible to estimate that about 5,000 stages would be necessary for one type of diffusion system and that a total area of many acres of diffusion barrier would be required in a plant separating a kilogram of U-235 each day. Corresponding cost estimates were tens of millions of dollars. For the centrifuge the number of stages would be smaller, but it was predicted that a similar production by centrifuges would require 22,000 separately driven, extremely high-speed centrifuges, each three feet in length at a comparable cost.

4.35. Of course, the cost estimates could not be made accurately since the technological problems were almost completely unsolved, but these estimates as to size and cost of plant did serve to emphasize the magnitude of the undertaking.


4.36. In September 1940, P. H. Abelson submitted to Briggs a 17 -page memorandum suggesting the possibility of separating the isotopes of uranium by thermal diffusion in liquid uranium hexafluoride. R. Gunn of the Naval Research Laboratory was also much interested in the uranium problem and was appointed a member of the Uranium Committee when it was reorganized under the NDRC in the summer of 1940. As a result of Abelson's suggestion and Gunn's interest, work was started on thermal diffusion at the National Bureau of Standards. This work was financed by funds from the Navy Department and in 1940 was transferred to the Naval Research Laboratory, still under the direction of Abelson, where it was continued.

4.37. We shall discuss the thermal-diffusion work further in a later chapter, but we may mention here that significant results had already been obtained by the end of 1941 and that in January 1942, using a single separation column, a separation factor had been obtained which was comparable or superior to the one obtained up to that time in preliminary tests on the diffusion and centrifuge methods.


4.38. It was pointed out in Chapter II that deuterium appeared very promising as a moderator because of its low absorption and good slowing-down property but unpromising because of its scarcity. Interest in a deuterium moderator was stimulated by experimental results obtained in Berkeley demonstrating that the deuterium absorption cross section for neutrons was, in fact, almost zero. Since oxygen has a very low absorption coefficient for neutrons, it was usually assumed that the deuterium would be used combined with oxygen, that is, in the very convenient material: heavy water. Work at Columbia on possible methods of large-scale concentration of heavy water was initiated in February 1941 under the direction of H. C. Urey (under an OSRD contract). Early in 1941, R. H. Fowler of England reported the interest of the British group in a moderator of deuterium in the form of heavy water and their conviction that a chain reaction would go in relatively small units of uranium and heavy water.

4.39. Urey and A. von Grosse had already been considering the concentration of heavy water by means of a catalytic exchange reaction between hydrogen gas and liquid water. This process depends on the fact that, when isotopic equilibrium is established between hydrogen gas and water, the water contains from three to four times as great a concentration of deuterium as does the hydrogen gas. During 1941, this exchange reaction between water and hydrogen was investigated at Columbia and in the Frick Chemical Laboratory at Princeton and extensive work was done toward developing large-scale methods of producing materials suitable for catalyzing the reaction.

4.40. The further development of this work and of other methods of producing heavy water are discussed in Chapter IX. Like the other isotope-separation work at Columbia, this work was relatively unaffected by the reorganization in December 1941. It is mentioned in preliminary fashion here to indicate that all the principal lines of approach were under investigation in 1941.


4.41. By the end of 1941 not very much progress had been made in the production of materials for use in a chain-reacting system. The National Bureau of Standards and the Columbia group were in contact with the Metal Hydrides Company of Beverly, Massachusetts. This company was producing some uranium in powdered form, but efforts to increase its production and to melt the powdered metal into solid ingots had not been very successful.

4.42. Similarly, no satisfactory arrangement had been made for obtaining large amounts of highly purified graphite. The graphite in use at Columbia had been obtained from the U. S. Graphite Company of Saginaw, Michigan. It was of high purity for a commercial product, but it did contain about one part in 500,000 of boron, which was undesirable.

4.43. Largely through the interest of Allison the possibility of increasing the production of beryllium had been investigated to the extent of ascertaining that it would be difficult and expensive, but probably possible.

4.44. Though little progress had been made on procurement, much progress had been made on analysis. The development of sufficiently accurate methods of chemical analysis of the materials used has been a problem of the first magnitude throughout the history of the project, although sometimes overshadowed by the more spectacular problems encountered. During this period C. J. Rodden and others at the National Bureau of Standards were principally responsible for analyses; H. T. Beans of Columbia also cooperated. By 1942 several other groups had started analytical sections which have been continuously active ever since.

4.45. To summarize, by the end of 1941 there was no evidence that procurement of materials in sufficient quantity and purity was impossible, but the problems were far from solved.


4.46. Prior to the autumn of 1941 there had been some exchange of reports with the British and some discussion with British scientific representatives who were here on other business. In September 1941, it was decided that Pegram and Urey should get first-hand information by a trip to England. They completed their trip in the first week of December 1941.

4.47. In general, work in England had been following much the same lines as in this country. As to the chain-reaction problem, their attention had focused on heavy water as a moderator rather than graphite; as to isotope separation, they had done extensive work on the diffusion process including the general theory of cascades. Actually the principal importance of this visit and other interchanges during the summer of 1941 lay not in accurate scientific data but in the general scientific impressions. The British, particularly J. Chadwick, were convinced that a U-235 chain reaction could be achieved. They knew that several kilograms of heavy water a day were being produced in Norway, and that Germany had ordered considerable quantities of paraffin to be made using heavy hydrogen; it was difficult to imagine a use for these materials other than in work on the uranium problem. They feared that if the Germans got atomic bombs before the Allies did, the war might be over in a few weeks. The sense of urgency which Pegram and Urey brought back with them was of great importance.


4.48. The appointment of a National Academy committee was mentioned in Chapter III. The committee's first report in May 1941 mentioned (a) radioactive poisons, (b) atomic power, and (c) atomic bombs, but the emphasis was on power. The second report stressed the importance of the new results on plutonium, but was not specific about the military uses to which the fission process might be put. Both these reports urged that the project be pushed more vigorously.

4.49. The third report (November 6, 1941) was specifically concerned with the "possibilities of an explosive fission reaction with U-235." Although neither of the first two National Academy reports indicated that uranium would be likely to be of decisive importance in the present war, this possibility was emphasized in the third report. We can do no better than quote portions of this report.

"Since our last report, the progress toward separation of the isotopes of uranium has been such as to make urgent a consideration of (1) the probability of success in the attempt to produce a fission bomb, (2) the destructive effect to be expected from such a bomb, (3) the anticipated time before its development can be completed and production be underway, and (4) a preliminary estimate of the costs involved."

"1. Conditions for a fission bomb. A fission bomb of superlatively destructive power will result from bringing quickly together a sufficient mass of element U-235. This seems to be as sure as any untried prediction based upon theory and experiment can be. Our calculations indicate further that the required masses can be brought together quickly enough for the reaction to become efficient...

"2. Destructive effect of fission bombs. (a) Mass of the bomb. The mass oj U-235 required to produce explosive fission under appropriate conditions can hardly be less than 2 kg nor greater than 100 kg. These wide limits reflect chiefly the experimental uncertainty in the capture cross section of U-235 for fast neutrons. . . (b) Energy released by explosive fission. Calculations for the case of masses properly located at the initial instant indicate that between 1 and 5 per cent of the fission energy of the uranium should be released at a fission explosion. This means from 2 to 10 × 108 kilocalories per kg of uranium 235. The available explosive energy per kg of uranium is thus equivalent to about 300 tons of TNT.

"3. Time required for development and production of the necessary U-235. (a) Amount of uranium needed. Since the destructiveness of present bombs is already an important factor in warfare, it is evident that, if the destructiveness of the bombs is thus increased 10,000-fold, they should become of decisive importance. "The amount of uranium required will, nevertheless, be large. If the estimate is correct that 500,000 tons of TNT bombs would be required to devastate Germany's military and industrial objectives, from 1 to 10 tons of U-235 will be required to do the same job.

"(b) Separation of U-235. The separation of the isotopes of uranium can be done in the necessary amounts. Several methods are under development, at least two of which seem definitely adequate, and are approaching the stage of practical test. These are the methods of the centrifuge and of diffusion through porous barriers. Other methods are being investigated or need study which may ultimately prove superior, but are now farther from the engineering stage.

"(c) Time required for production of fission bombs. An estimate of time required for development, engineering and production of fission bombs can be made only very roughly at this time. "If all possible effort is spent on the program, one might however expect fission bombs to be available in significant quantity within three or four years.

"4. Rough estimate of costs. (The figures given in the Academy report under this heading were recognized as only rough estimates since the scientific and engineering data to make them more precise were not available. They showed only that the undertaking would be enormously expensive but still in line with other war expenditures.)"

4.50. The report then goes on to consider immediate requirements and desirable reorganization.


4.51. At the end of Chapter I we summarized the knowledge of nuclear fission as of June 1940, and in Chapter II we stated the outstanding problems as of the same date. In the light of these statements we wish to review the eighteen months' progress that has just been recounted. The tangible progress was not great. No chain reaction had been achieved; no appreciable amount of U-235 had been separated from U-238; only minute amounts of Pu-239 had been produced; the production of large quantities of uranium metal, heavy water, beryllium, and pure graphite was still largely in the discussion stage. But there had been progress. Constants were better known; calculations had been checked and extended; guesses as to the existence and nuclear properties of Pu-239 had been verified. Some study had been made of engineering problems, process effectiveness, costs, and time schedules. Most important of all, the critical size of the bomb had been shown to be almost certainly within practical limits. Altogether the likelihood that the problems might be solved seemed greater in every case than it had in 1940. Perhaps more important than the actual change was the psychological change. Possibly Wigner, Szilard, and Fermi were no more thoroughly convinced that atomic bombs were possible than they had been in 1940, but many other people had become familiar with the idea and its possible consequences. Apparently, the British and the Germans, both grimly at war, thought the problem worth undertaking. Furthermore, the whole national psychology had changed. Although the attack at Pearl Harbor was yet to come, the impending threat of war was much more keenly felt than before, and expenditures of effort and money that would have seemed enormous in 1940 were considered obviously necessary precautions in December 1941. Thus it was not surprising that Bush and his associates felt it was time to push the uranium project vigorously. For this purpose, there was created an entirely, new administrative organization which will be described in the next chapter.