37. Letter From Secretary of Defense McNamara to the President’s Special Assistant for National Security Affairs (Bundy)1

Dear Mac:

Glenn Seaborg’s letter of November 23, 19642 raises basic issues about that aspect of U.S. policy against proliferation of nuclear weapons and delivery systems which is represented by NSAM 294.

I cannot agree with the position taken by AEC on the export of certain advanced computers and enriched U-235 to France. It suggests that NSAM 294 and over-all U.S. policy calls for the denial of assistance to France only if each denial action involves an item of equipment or technology that is in itself a controlling, critical or essential item in the French program. This position, in my judgment, is unrealistic and self-defeating if the U.S. seriously intends to follow the intent of NSAM 294 “not to contribute or assist” in the development of national nuclear capabilities.

We all recognize, of course, that France or any other technologically advanced nation has the capability to develop nuclear weapons and delivery systems, and that the impact of U.S. denial of equipment and technology may in many cases not be determinative. Nevertheless, export controls, as one of several means of carrying out the anti-proliferation policy, will—by cumulative effect—retard, increase the cost of, or discourage French and other national efforts. In my judgment, therefore, the U.S. should be firm and consistent in denying whatever would measurably assist the program of any nation that aspires to a nuclear weapons capability, unless such assistance would be clearly insignificant. The approach taken by the AEC would, I suggest, unnecessarily weaken such an export control policy and make that policy inconsistent with the over-all U.S. anti-proliferation objective.


The Computer Cases. I disagree with the AEC recommendation that export be permitted. These computers are designed admirably for atomic research, and provide greatly enhanced ability for development of French nuclear weapons and strategic delivery systems (although they of course have other uses). One computer, the IBM 360-92, will be the largest and most advanced model ever produced. Its acquisition by France in about two years certainly would be an important, though [Page 62] perhaps not an immediately decisive factor in the long-range French nuclear weapons effort. The CDC 3600, the second computer, is one of the most advanced currently in use. Its proposed delivery directly to a French nuclear weapons facility convinces me that export should be denied.

I recommend, therefore, that the U.S. deny the export of these computers under NSAM 294 and export of any others that would make a similar significant contribution to nuclear weapons or delivery systems development. The only condition under which I would agree to approve such items for export is a firm assurance by the French Government that they would not be so used.

I have the following additional comments on the AEC memorandum attached to Glenn Seaborg’s letter:

Paragraph C (Memo, p. 1) suggests that the advanced computers are not absolutely required by the French, pointing to the fact that a U.S. thermonuclear device was developed without such a computer capability. This is true, but our first device was very awkward and heavy, weighing 42,000 lbs. The French can develop thermonuclear explosives of this type without the very large advanced computers involved here, but they cannot design a small deliverable warhead. We do not want to provide them the ability in the next few years to reduce, say, a 10,000 lb thermonuclear device to a missile warhead of 750 lbs. This difference is vital.
Paragraph C also highlights what most disturbs me about the AEC view of NSAM 294. The question is not whether the French are “dependent upon” the item sought to be exported (i.e., “but for” securing that item their program will founder), but rather whether the item will significantly assist their program. The AEC apparently does not dispute that the computers involved meet the latter standard. Indeed, both DOD and AEC today rely heavily upon comparable equipment in U.S. nuclear weapons and delivery system programs, the AEC military uses being interchangeable with peaceful uses.
The acknowledgment in paragraph D that these computers would assist the French program “after 1969” should in itself be sufficient to require denial under NSAM 294. Had the U.S. been foresighted enough to act more energetically only a few years ago along the lines NSAM 294 now calls for, we might today be confronted with far fewer problems stemming from French nuclear proliferation.
The statement is made at page 2 that the French would construe denial to mean that the U.S. is refusing “to cooperate with them in an area of fundamental and unclassified technology.” Of course, the computers are “unclassified” but they will be used for highly classified nuclear weapons research programs, and can provide a significantly improved base for strategic delivery system development as well [Page 63] (submarine hull design, missile performance simulation, etc.). Admittedly, the French may be unhappy, and their reaction will be understandable. This is a risk, and a predictable result of applying NSAM 294. The real issue is whether or not the U.S. seriously intends to retard the French program, knowing that they will disapprove of our actions. Moreover, I question whether the French adverse reaction is not being stressed too much. They have known of our anti-proliferation policy for some time now.
At page 2, an inordinate stress is placed on the dollar value of the sales involved. The dollar value for export of the many, many types of computers (about 150) not falling within NSAM 294 is far greater than the value of the small number of types of advanced computers (about 12) comparable to those at issue here. Obviously I do not oppose the export of the 150 other types which have many research, industrial, and commercial applications (as well as substantial utility in nuclear weapons programs). As a rough estimate, these 150 other types comprise about 95% of the dollar value of the total market.
The reference at page 2 to the Ferranti capability is probably incorrect in part. Information from the CIA staff indicates that neither this U.K. firm, nor the company to which it apparently recently sold its large-computer production facilities (ICT), plans to produce a computer comparable to the IBM 360-92. As a result, the largest computer developed or known to be under development outside the U.S. has been estimated to have only 5%-15% of the capability of the 360-92. Even this limited capability may be important enough to call for discussion with the source country (U.K.) to influence it to withhold any such computers under an anti-proliferation policy akin to our own. The results of such efforts should not be prejudged.
I question the suggestion at page 2, third paragraph, that existing French computer capability devoted to non-weapons research could be harnessed to assist significantly their nuclear weapons effort, thereby obviating the need for the advanced computers at issue. Pooling a number of large computers does not provide a substitute for a single advanced machine such as the IBM 360-92. Certain sophisticated problems of extreme size and complexity do not lend themselves to piecemeal solution; in other instances the time required for solution makes pooling unrealistic. For the volume of work which could be processed, scheduling and central programming would be quite difficult; there would be inefficient fragmenting of many computing jobs; and the computer time required for much of the necessary work would be increased. Further, diverting existing French computers into the pool would starve other segments of their industry. Even with this pooling effort, the French nuclear weapons program would in all probability be held back; at least it would not be speeded up, as it would be with the advanced computers AEC would permit to be exported.

The Enriched U-235 Case. An initial DOD position was furnished on November 6, 1964 to AEC by Mr. W.J. Howard, Assistant to the Secretary of Defense (Atomic Energy). (A copy is attached.)3 The position is basically that the export license should be denied because the material is expressly for the French land-based prototype submarine nuclear propulsion plant, and the French intend that submarine to be part of a missile delivery system. The 1959 Agreement under which earlier shipments were made is relevant, but not determinative. Present conditions, especially the current U.S.-French-NATO nuclear controversies, are very different from those prevailing in 1959. Subsequent to Mr. Howard’s letter, [6 lines of source text not declassified]4

For the above reasons, I reaffirm the stated DOD position in opposition to the export and would be most disinclined to certify, as the 1959 Agreement requires, that the transfer of enriched uranium to France at this time will promote, and will not constitute an unreasonable risk to U.S. defense and security. As Mr. Howard’s letter of November 6 indicates, there are adequate grounds under the 1959 Agreement to support denial.

In connection with the U-235 export, I note that the AEC memorandum (page 4, paragraph D) cites as a reason for favorable action the fact that denial would have only a short-range (“limited to the next two years”) effect. In the computer cases, the AEC spoke in favor of export because denial would have only a long-range (post-1969) effect. Somewhat questionably both grounds are urged concurrently in support of permitting the exports.

In conclusion, I am hopeful that the differences of interpretation which the AEC letter has usefully brought to your attention regarding the computer and U-235 transactions can be quickly resolved. This should eliminate a major obstacle to implementing the NSAM 294 policy effectively.


Bob 5
  1. Source: National Archives and Records Administration, RG 59, S/S-NSC Files: Lot 72 D 316, NSAM 294. Secret. Copies were sent to the Secretary of State and to the Chairman of the Atomic Energy Commission.
  2. See Document 35.
  3. Not found.
  4. [Text not declassified]
  5. Printed from a copy that bears this stamped signature.