[Page 447]

188. Memorandum from Secretary of Defense Brown to the President’s Assistant for National Security Affairs (Brzezinski)1


  • Nuclear Weapons Stockpile Reliability Under a CTB (U)

(S) Almost two months ago, I raised the issue of what nuclear experiments would be permitted under a CTB. Since then we have analyzed this issue in some detail.

(S) The purpose of this paper is to discuss the issue of maintaining confidence in the nuclear stockpile under a CTB. Specifically, this paper will address: the ways we have ensured stockpile reliability in the past; the potential problems we will face under a CTB; the alternatives available to maintain high confidence in stockpile reliability in the future; and the implications of these alternatives in terms of verification, our non-proliferation objectives, U.S.–U.S.S.R. asymmetries, and achieving Congressional ratification of the treaty.

(S) I begin with the assumption that without question it will be U.S. national policy under a CTB to maintain high confidence in our nuclear deterrent forces. An essential element of this will be confidence in the reliability of our nuclear weapons stockpile.

(S/RD/N) Maintenance of Stockpile Reliability in the Past. In the past, we have ensured reliability of the nuclear weapons in our stockpile by a comprehensive program in which nuclear testing played a key role at several points.2 First, during development, each device was extensively tested in a program that typically started at low yields for development of the primary, and progressed up to full yield detonations and nuclear effects tests of the weaponized version. Second, during the stockpile life of the weapon, reliability problems uncovered in the stockpile surveillance program were resolved by expert design teams who were familiar with the results of all past testing of the weapon and whose competence was maintained by participation in a continuing nuclear test [Page 448]program. When necessary, individual reliability problems were investigated and resolved by specific nuclear tests. The adequacy of fixes or design modifications occasionally had to be verified by nuclear testing. Finally, the weapon was replaced in the inventory well before the end of its nominal service life by a new design which was perfected through its own development test series.

(S/RD/N) The yield levels of the above test programs varied, of course, from a few pounds to full weapon yield. [5 lines not declassified]

(S/FRD) Potential Problems Under a CTB . Under a CTB—whatever its terms—we will not be able to continue maintaining the stockpile as before. The outlook for stockpile reliability depends upon the type of CTB we aim for and achieve.

(S/FRD) If a CTB allowed no weapons testing having any nuclear yield, our confidence in the reliability of the stockpile would remain high initially, and would then decline at an accelerating rate, possibly reaching quite low values. Methods to quantify this degradation are not now known. The reasons for its occurrence are complex. In the worst cases, a specific problem could require retirement of entire major systems. As warheads reach the end of their service life—and some in the stockpile will reach that point relatively soon—we are not certain that we can replace them with exact copies in which we have high confidence. There are a number of reasons for this, including our inability to reproduce materials or duplicate manufacturing processes exactly; some of these may be prohibited because of present day environmental and safety regulations. Finally, as delivery systems are necessarily changed with advancing technology or new missions and threats, we are not certain we can adapt existing warheads with high confidence. The possibility of such situations arising under a CTB prompts consideration of alternative CTB concepts.

(S) The central question to be asked with regard to all alternatives to an ideal “zero-yield” treaty is: Which alternatives would maintain the highest stockpile reliability over the long term without jeopardizing the potential benefits, particularly the non-proliferation benefits, of a CTB?

(S/FRD) Alternative Approaches. Three alternative approaches appear worthy of consideration, either singly or in combination. We would argue that each could be made compatible with the concept of a CTB because of the special requirements of nuclear weapon states (NWS) to ensure the reliability of their stockpiles. In brief form, these alternatives are:

—(S/FRD) Permitted Experiments. Allow low-yield (kiloton or sub-kiloton) experiments below some designated threshold with the purpose of contributing to stockpile maintenance.

[Page 449]

—(S/FRD) Quota. Permit some low number of nuclear tests (e.g., 2–5) annually for the purpose of stockpile maintenance.

—(S/FRD) Phase-out. Provide a transition period (e.g., 10 years) during which testing, within some limits to be determined, would be allowed for the purpose of adapting existing stockpiles to lower-yield test conditions.

(S/RD/N) Permitted Experiments (U).

(S/FRD) Provisions for “permitted experiments” would allow the use of explosive release of nuclear energy for maintaining our confidence in stockpile reliability. This alternative would make a clear-cut distinction between “nuclear testing” which would be prohibited and “nuclear experiments” which would be permitted. Energy-related work, such as inertial confinement fusion, would continue.

(S/FRD) The value of low-yield nuclear experiments for maintenance of confidence in the nuclear weapons stockpile would vary with the yield allowed. Such experiments may range from the minimum “observable” nuclear yield, about [numbers not declassified] kiloton, to an upper limit of a few kilotons. They are different in a most essential way from zero-yield hydrodynamic testing in which no fissile material is involved. [4 lines not declassified]

[6 paragraphs (44 lines) not declassified]

(S/RD/N) [3 lines not declassified] However, testing above 1 KT, except perhaps in connection with a quota or phase-out provision, may not be considered compatible with the concept of a CTB.

(U) Another way to summarize the requirements for maintenance of stockpile reliability and to show the yield dependencies of other related aspects of a CTB is shown in Table 1.3

(S) There are at least two ways that “permitted experiments” could be incorporated in a CTB. One approach would be to ban “nuclear explosions” in the treaty, and then define the term in such a way as to exclude the tests we wish to allow. A possible formulation might be: “For the purposes of this treaty, a nuclear explosion is defined as an event producing more thancalories yield from a super-critical assembly in less than a few microseconds.” A second approach would be to ban nuclear explosions (undefined) and then explicitly identify an exclusion. There are a number of ways this exclusion could be worded; for example: “The prohibition of Articledoes not apply to fully contained nuclear experiments conducted in reusable laboratory facilities.” This second approach, however, would be applicable, because of [Page 450]technical limits on containment, only to nuclear experiments whose yield limits would be about 100 tons.

(S) Quota and Phase-out Alternatives. Both the quota and phase-out alternatives are variants on the permitted experiments theme. However, as presently conceptualized, both approaches would permit testing at a much higher kiloton threshold than envisaged under the permitted experiments option.

(S) The quota approach would apply a stringent limit on the number of tests allowed each year (e.g., 2–5). The purpose of the testing would be stockpile maintenance, i.e., tests only for conversion to “woodenized” warheads, recertifying rebuilt warheads, or fixing problems. If no lower limit were imposed, the 150 KT threshold of the TTBT could govern, but this approach could be coupled with a reduction of the 150 KT threshold.

(S) The phase-out approach would be constructed explicitly to attempt transition to “wooden” warheads which might be designed to be replaceable with only limited design guidance or testing. As in the alternative above, the TTBT could govern yield unless otherwise specified. The duration could be specified by the number of years or tests or both (e.g., 10 years or 25 tests, whichever comes first).

(S/RD) Combination of Alternatives. Possibly more attractive than any of the individual alternatives—both for achieving the “tightest” possible CTB for non-proliferation purposes and for achieving high stockpile reliability—would be a combination. For example, a quota (5 tests per year) at a reduced yield threshold (15 KT) for a phase-out period (5 years), followed by permitted experiments (sub-kiloton), and with the purpose of all tests being rigorously limited by description to stockpile maintenance. During the phase-out period, a periodic review would be conducted to assess the progress in achieving the desired goal. If success was not likely, it might not follow that the transition would be made to the “permitted experiments” phase. If it made negotiating strategy easier, this could be approached on a “Phase I, Phase II” basis. In all alternatives nuclear effects testing could be “piggy-backed” on weapons tests. (Note that this combination will not allow high-yield testing to resolve problems that may occur with secondaries.)

(S/FRD) Implications of the Alternatives (U)

[1 paragraph (10 lines) not declassified]

(S) Non-Proliferation. The implications of the various alternatives to maintain stockpile reliability under a CTB for our non-proliferation objectives are difficult to determine. By agreeing to a CTB, the U.S., U.S.S.R., and U.K. will have taken a major step toward achieving the over-all objectives of the Non-Proliferation Treaty. We will be severely constrained in the development of new weapons even if nuclear experi[Page 451]ments, a quota on testing, or a phase-out of testing are permitted (for nuclear weapons states only). At issue here in relation to our NPT objectives is how a CTB which allowed (or did not allow) one or more of the various alternatives would affect the nuclear development decisions of those states who have not signed the NPT.

(S) Any exceptions for nuclear weapons states will be characterized by some of the non-nuclear states as discriminatory. States which want a rationale for continued non-participation in the NPT are likely to exploit any exception from an ideal zero-yield treaty. On the other hand, I see no reason to believe that a zero-level treaty (as opposed to one allowing experiments, a quota, or a phase-out) will provide sufficient reason for those same states to participate in the NPT. The motivations behind the decisions for nuclear weapons are far more complex, and virtually impossible to characterize in advance. We should distinguish in our own thinking between those real motivations, on the one hand, and excuses on the other.

(S/RD) U.S.–U.S.S.R. Asymmetries and Congressional Ratification. A third consideration is how any of the various alternatives for maintaining stockpile reliability under a CTB would affect asymmetries in the reliability of U.S. and U.S.S.R. stockpiles. Although much is unknown in this area, several factors appear to give the Soviets an advantage in reliability of their stockpile. First and most important, they can retain their weapon design lab personnel for many years simply by control of personnel assignments. Second, their missiles have a larger throw-weight by a significant amount, giving them more flexibility in correcting problems without nuclear testing (i.e., by falling back to simpler, heavier warhead designs).

[1 paragraph (6 lines) not declassified]

[8 lines not declassified] Higher yield primaries, while almost surely weighing more, would be less subject to deleterious effects of small changes in production processes, materials, etc., and thus could have better long-term reliability.

(S) It would appear, therefore, that the Soviets would begin a CTB with a “built-in” asymmetry in their favor. The quota and phase-out alternatives are the ones that would contribute more directly to minimizing the effects of this asymmetry.

(S) With respect to the permitted experiments alternative, the question of potential asymmetries becomes intertwined with the verification capabilities of both sides. [5 lines not declassified] A unilateral U.S. failure to carry out experiments at that level could produce significant technical asymmetries. It would almost certainly have a considerable political impact if it were believed that we were not, but the Soviets were, doing such experiments.

[Page 452]

(S) Another consideration is that we should expect close Congressional scrutiny of the stockpile reliability and verification problems associated with a CTB. A treaty which constrains experiments to the kiloton and sub-kiloton range—let alone a still lower level—is likely to be attacked by some as a “give away” that concedes a permanent reliability advantage to Soviet nuclear forces. On the other hand, a CTB treaty that has a higher yield quota or phase-out provision will probably be attacked by others as a significant departure from the ideal of a “comprehensive” test ban.

(S/FRD) Options (U)

(S) At issue is whether some form of limited nuclear testing—limited in yield, numbers of tests, or duration—should be allowed under a CTB in order to maintain, over the long term, high confidence in the reliability of our nuclear weapons stockpile.

(S) At this time, there is no known method of ensuring high stockpile reliability that does not include some nuclear testing. Yet any proposal for modifying a “zero-yield” CTB would conflict with the basic purposes of an ideal CTB, the degree dependent upon the type of modification desired. Weighing together these two policy requirements is the tough problem.

(S) The options which appear to offer the best approach toward satisfactorily gaining both objectives of stockpile reliability and non-proliferation are listed below:

1. (S/FRD) Phase-out, i.e., some form of quota testing, restricted solely to nuclear weapon states, at a limit of 15 KT or less, over a specified period of time. [2 lines not declassified] As to the length of the transitional period, it is again noted that methods to quantify confidence levels in the reliability of our stockpile, in a condition of non-testing, are not known. Sufficient time to determine the feasibility of a “woodenized” stockpile and to permit an orderly transition to a posture that would provide adequate assurance of high confidence in stockpile reliability under a CTB would be essential criteria. For a short transition period, e.g., 5 years or less, a review provision following it would be desirable.

2. (S/FRD) Small quota of tests per year. A quota of perhaps two tests per year at some fixed yield limit, perhaps as low as 5 KT or even as high as the 150 KT TTBT limit, might provide sufficient opportunity for assuring stockpile confidence. This rate would be so much lower than the current testing rates of both the U.S. and the U.S.S.R. that it should allay most concerns about the development of new weapons. However, this approach would be a major deviation from a pure CTB. To minimize its impact on our non-proliferation goals, it would require full clarification to the non-nuclear weapon states of the need of nuclear [Page 453]weapons states to maintain a reliable nuclear weapon stockpile. The advantage of this approach is that throughout the yield range it continues testing in the mode where our past experience lies, and also [less than 1 line not declassified]

(S/FRD) Under both Options 1 and 2, sub-kiloton nuclear experiments, e.g., 100 tons, would be allowed indefinitely. They would assist in maintaining high levels of scientific expertise at the laboratories. This would also allow energy-related work to go on, and would accept the fact that military benefits, e.g., nuclear weapons effects testing, cannot be excluded from such work.

3. (S/RD) Permitted Experiments [less than 1 line not declassified]. Stockpile confidence would be lower than under options 1 and 2 above. While testing at levels less than about [less than 1 line not declassified] would not contribute directly to stockpile reliability, the positive benefit of such testing would be in encouraging the retention of laboratory expertise which would help indirectly to assure stockpile confidence. Full support by the government in retaining the high scientific quality of the laboratories would be essential. [2 lines not declassified]

(S) There are other variations which could also contribute to stockpile reliability but would probably contribute less toward non-proliferation objectives than the above options.

1. (S/FRD) [2 lines not declassified] Such tests would contribute directly to stockpile reliability but may not solve all stockpile maintenance questions requiring testing. [1 line not declassified]

2. (S) Formal Review. Under this variant, there would be no provision for nuclear tests for stockpile reliability during the period of the agreement; however nuclear experiments under 1 kiloton would be permitted. The agreement would include a provision for a formal review at the end of the 3 to 5 year initial duration period. At that time, each party would have the right to conduct those nuclear tests at a specified yield which are certified by each government to be required to maintain stockpile reliability.

(S/FRD) Recommendation. The issues identified in this paper concerning some form of limited testing under a CTB for maintaining high confidence in the reliability of our nuclear stockpile are of sufficient national importance to warrant an SCC meeting to obtain a conclusive decision.

Harold Brown
  1. Source: Washington National Records Center, OSD Files: FRC 330–81–0202, Box 53, A–400.112 TEST BAN (1 Jan–15 Mar) 1978. Secret; Restricted Data.
  2. The U.S. stockpile consists primarily of high-technology warheads, each uniquely optimized for its intended delivery system and its military mission. Each type warhead is custom-designed according to the required yield, output, weight, size, and shape. Usually, old designs are not rebuilt for new applications. Maximum emphasis is placed on reduced size and weight, economical use of special nuclear materials (oralloy, plutonium, and tritium), maximum yield-to-weight ratio, inherent one-point safety, security, and survivability. [Footnote is in the original.]
  3. [Table 1 not declassified]