the two source types (Ref 13); (2) earthquake sources, because of a larger relaxation volume, are more energetic at long periods (20 seconds), relative to short periods (1 second), than explosions are.

Determination of earthquake source depth

Several refinements have been made (Ref 14) to the technique of using the spectrum of the spectrum, or cepstrum, to detect the delay times of surface reflections found in the short period coda of seismic signals from shallow sources. These delay times are directly proportional to the depth of the source. Hence, their accurate determination is important in seismic verification. The delay times may be directly measured from isolated peaks or maxima in the cepstrum. In order to apply this technique to data from a network of stations, a single network cepstrum is generated by randomly delaying and summing the individual cepstrum from each network site. That combination of delays which yields to sharpest peaks in the summed cepstrum is taken to be the correct one. Strictly speaking this technique is only applicable when all of the recording stations are about equidistant from the source. The combined cepstrum is then searched using theoretical patterns which contain the effects of both compressional and shear wave reflections. This matched filter or correlation technique attempts to collect all the source depth information from the short period coda onto a single trace containing a single or predominant maxima. Initial results using these techniques are encouraging, particularly for events at depths of many shallow earthquakes, 20-100 km.

SECTION IV

CONCLUSION

We have outlined above the remaining technical problems associated with seismic verification of a comprehensive test ban treaty as we saw them at our last presentation to this Committee. We have described the data resources which we are developing in an effort to gain solutions to these problems. The new data is now becoming available and is being distributed to research seismologists. We have set down the results of our evaluation of our current data systems and their capabilities. We have brought out the more interesting new results of our ongoing research effort in this area. We now look forward to applying the newly completed data sources and facilities to these new ideas and concepts, and to the verification of potential capabilities listed in the last section.

REFERENCES

1. Sorrells, G. G., "A Preliminary Investigation into the Relationship between Long-Period Seismic Noise and Local Fluctuations in the Atmospheric Pressure Field," Geophys. J. Roy astr. Soc., Vol. 26,

1971.

2. Ziolkowski, A., "Prediction and Suppression of Long-Period NonPropagating Seismic Noise," Bull. Seis. Soc. Am., Vol. 63, June

1973.

3. Lacoss, R. T., R. E. Needham, and B. R. Julian, "International Seismic Month Event List," M.I.T Lincoln Laboratory Technical Note 1974-14, 27 February 1974.

4. Needham, R., "Worldwide Detection Capability of a Prototype Network of Seismograph Stations," M.I.T. Lincoln Laboratory Technical Note 1975-42, 19 August 1975.

5. Filson, J. R., "Long Period Results from the International Seismic Month," M.I.T. Lincoln Laboratory Technical Note 1974-15, 4 March 1974.

6. Basham, P. W., and F. M. Anglin, "Multiple Discriminant Screening Procedure for Test Ban Verification," Nature, Vol. 256, December 1973.

7. Lambert, D. G., A.I. Tolstoy, and E. S. Becker, "Seismic Detection. and Discrimination Capabilities of the Very Long Period Experiment-Final Report," Texas Instruments Report No. TR-74-07, 9 December 1974.

8. Evernden, J. F., "Further Studies in Seismic Discrimination," Bull. Seis. Soc. Am., Vol. 65, April 1975.

9. Evernden, J. F., "Study of Seismological Evasion," Bull. Seis. Soc. Am., Vol. 66, (Parts I and II-February 1976, Part III— April 1976).

10. Forsyth, D., "Higher Mode Rayleigh Waves as an Aid to Seismic Discrimination," Bull. Seis. Soc. Am., in press.

11. Tatham, R., D. Forsyth, and L. Sykes, "The Occurrence of Anomalous Seismic Events in Eastern Tibet," Geophys. J. Roy. astr. Soc., in press.

12. Archambeau, C. B., "Investigations of Tectonic Stress," Final Technical Report 1 November 1973-31 August 1975, University of Colorado, 15 August 1975.

13. Gilbert, J. F., "The Relative Efficiency of Earthquakes and Explosions in Exciting Surface Waves and Body Waves," Geophys. J. Roy. astr. Soc., Vol. 33, 1973.

14. Page, E., and R. Bauman, "Determination of Seismic Source. Depths from Differential Travel Times," Final Report, Ensco, Inc., July 1975.

Statement by the British Expert (Fakley) at an Informal Meeting of the Conference of the Committee on Disarmament: Comprehensive Test Ban, April 20, 19761

NUCLEAR TEST BANS

1. As has already been said this afternoon there are two principal technical difficulties standing in the way of successful negotiation of an agreement to ban the testing of nuclear weapons and other explosive nuclear devices in all environments; these are the question of

'The statement was subsequently circulated in the CCD by the U.K. delegation in a working paper, CCD/492.

verification and the problem of peaceful nuclear explosions. These two topics are bound to occupy a considerable proportion of the time available for discussion here this week.

2. There is a long history of debate on how it might be possible to verify by national technical means that the terms of any comprehensive nuclear test ban treaty were being observed. Verification that testing is not taking place in the environments proscribed by the 1963 Partial Test Ban Treaty 2 can be satisfactorily achieved and hence the current problem remains one of finding means whereby underground nuclear tests can be detected and identified at long range. To this end the United Kingdom, along with other countries, has been devoting considerable resources to seismic research. The results of this work have been presented to the Conference of the Committee on Disarmament from time to time and my delegation has made a review of the United Kingdom contributions for this meeting. Our broad technical assessment is that it is now possible to establish a world-wide network of seismic stations capable of detecting and identifying seismic events down to a body wave magnitude of about 42. At somewhat lower magnitudes, events would still be detected but, seismologically, it would no longer be possible in many cases to decide with adequate assurance whether or not a particular event had been caused by an earthquake or an explosion. At even lower magnitudes, events could not even be detected with sufficient probability against the background seismic noise.

3. We equate a seismic magnitude of m, 41⁄2 with an explosive yield of about 5 kilotons when the explosive is close coupled in granite or other hard rock. However, anyone seeking to evade a verification system established to monitor a comprehensive test ban is not going to set off his explosive in hard rock with close coupling if he can possibly avoid it. At a very minimum he would choose a more favourable geologic environment. Experimental results published in the United States show that explosive devices detonated in a suitable thickness of dry alluvium would produce seismic signals which were lower in magnitude by a factor of up to ten as compared with those from the same device detonated in hard rock. Therefore, in a place where there was a 300m or more thickness of dry soft rock, the detection and identification threshold of a verification system must currently be measured in terms of some tens of kilotons.

4. There are other ways of evading the verification system. One such way is the de-coupling of an explosion by setting it off in an underground cavity. Theoretical studies and small-scale experiments suggest that de-coupling factors of ten or more might be achieved by such means. For example it has been established that a cavity some 110 metres in diameter would be capable of de-coupling a 10 kiloton shot so that it would produce a seismic signal equal to that from a fraction of a kiloton shot close coupled in the same medium. Another way is to seek to hide the seismic signal from an explosion in the

2 For the treaty text, see Documents on Disarmament, 1963, pp. 291–293.

signal produced by a naturally occurring earthquake; this would require awaiting until a suitable earthquake was available to mask the explosion signal and clearly this would be a disadvantage.

5. We can argue indefinitely about the practicability and effectiveness of the various measures for reducing the efficiency of a verification system for monitoring a comprehensive test ban but we cannot afford to ignore them. Taking them into account in a general sense, it must be concluded that it is not possible to propose a seismic verification system capable of giving sufficient assurance that tests with yields up to a few tens of kilotons were not being carried out. This conclusion takes into account that to be reasonably sure of escaping detection, a would-be violator would have to be conservative in the choice of maximum yield he is prepared to test and would therefore have to leave some margin between this maximum and his assessment of the threshold capability of the verification system.

6. We therefore come to a crucial question. Given that tests of up to some tens of kilotons are not adequately verifiable, is this significant in a comprehensive test ban treaty context? As far as nuclear weapon states are concerned, nuclear testing is carried out for four main purposes. These are for:

(a) the maintenance of stockpiled weapons in an entirely safe yet serviceable condition;

(b) the investigation of the effects which nuclear explosions have on other items of equipment;

(c) the modification of nuclear explosive devices so that they can be fitted into new delivery systems;

(d) the study of the physical principles of nuclear devices.

Can any of these purposes be served by tests with yields of up to a few tens of kilotons?

7. The maintenance of stockpiled weapons in a safe yet serviceable condition is of vital importance whilst such weapons continue to exist. Like all other types of hardware, nuclear weapons age; in fact they tend to age more rapidly than most other weapons because of the characteristics of the special materials used in their construction. From time to time they have to be repaired and this frequently requires some of the components to be replaced. These replacements are likely to be somewhat different from the originals if only because available materials will have changed and because safety and security requirements have been made more stringent. The changes involved are usually small but it cannot be ascertained whether or not they are significant without nuclear testing. In order to be sure that the refurbished weapons remain safe and serviceable, it is necessary to carry out a full scale test but a reduced-scale test can offer the required assurances in some circumstances. A significant amount of stockpile weapons testing could be carried out at yields of a few tens of kilotons.

8. The role of nuclear testing to investigate the effects of nuclear explosions on other items of equipment is obvious. It is essential to assess the vulnerability of defence equipments to the nuclear explo

sions they may face in the course of their use. Some assessment work can be carried out theoretically and by using simulators but the final proof that equipment will withstand the effects of nuclear explosions to a sufficient degree can only be obtained by exposing it to the output from an actual nuclear explosion. Effects testing can be carried out using a source with a yield of a few tens of kilotons.

9. From time to time, requirements arise for new weapon systems to meet the changing strategic and tactical warfare situations and when any new delivery system is being developed, the problem of producing a suitable warhead has to be solved. The nuclear warhead has to be tailored to its delivery vehicle and to its operational role and in designing the warhead to achieve these objectives, uncertainties arise over whether the design incorporates changes from past knowledge which affect either its operation or its safety and security. These uncertainties can only be resolved finally by nuclear testing and normally such proof testing can be carried out if necessary at yields of a few tens of kilotons.

10. Finally, nuclear testing is still carried out to improve the understanding of the physical principles of nuclear weapon design. Many weapon physics experiments can be carried out at yields of some tens of kilotons and such experiments can lead to improvements in weapon design.

11. This summary of the current purposes of nuclear weapon testing shows that much could be achieved by tests which fall into a yield range below the existing estimates of the threshold of a verification system for a comprehensive nuclear test ban based on national technical means, even if these have a wide world distribution. On technical grounds the conclusion is simply that testing below the seismic verification threshold would be significant. Non-intrusive verification does not give the necessary assurance that significant nuclear testing is not taking place. A lack of confidence in the effectiveness of the verification system would lead to unacceptable uncertainty in the nuclear weapon states and could itself have a destabilizing effect irrespective of whether or not violations of a comprehensive test ban were taking place.

12. The development of an adequate national verification system for all underground nuclear tests remains an unresolved problem. The other major technical difficulty relates to so-called Peaceful Nuclear Explosions (PNEs). We have to ask the question, "How can one ensure that a nuclear explosion, which ostensibly is being carried out for peaceful purposes, does not give opportunities for gaining weaponsrelated benefits?" Of the types of weapon related tests already described, a PNE could most readily be used to check that a refurbished warhead was functioning correctly. No special arrangements would be needed to meet this objective; one would simply use a weapon from the stockpile as a peaceful nuclear explosive. The fact that, on detonation, the explosive gave the expected yield would confirm that the weapon was functioning as intended and this would relieve the tester of the uncertainty about the viability of his stockpile weapon. Even