The U.S. Department of Energy conducts a variety of "subcritical" nuclear weapons tests.  This type of test is termed "subcritical" because it uses fissile materials but there is no self-sustaining nuclear chain reaction.  Most subcritical tests employ weapons grade plutonium (Pu-239) in various configurations.  In recent years, subcritical tests have been conducted underground at the Nevada Test Site.  Subcritical tests also can be conducted using an isotope of plutonium with a higher critical mass (Pu-242) which allows test devices to more closely resemble actual nuclear weapons triggers in size and shape.  Both kinds of tests also can be conducted in some cases in vessels above ground.

When conducted underground at the same site used for full-scale nuclear weapons tests, subcritical experiments make verification of a test ban more difficult, and manifest to the world both the existence of a vigorous nuclear weapons research program and the intention to retain the capability for full-scale underground tests.  It is evident that subcritical testing programs conducted by the nuclear weapons states at nuclear test sites are likely to make verification of a test ban more difficult.  Suzanne Jones and Frank von Hippel wrote that:

"Seismic measurements can place an upper limit on the total explosive yield of a test, adequate to rule out the possibility that a full-scale nuclear test had been conducted. But seismic data would be of no use in determining what fraction of the energy from an explosion was nuclear.  If other countries wished to know whether a subcritical or hydronuclear experiment had taken place, how could they tell the difference?

Evidence that this question is not purely academic is provided by alleged activities at the Russian nuclear test site Novaya Zemlya, near the arctic circle in January 1996.  According to leaks to the Washington Times, intelligence information on these activities led some U.S. officials to suspect that a nuclear test had occurred.  One government official was quoted as saying that 'many Pentagon officials have few doubts and believe Moscow set off a small nuclear weapon.'  In the same article, however, State Department Spokesman David Leavy was quoted as saying, 'It is the view of the United States that the Russian moratorium on nuclear testing is continuing.'

The confusion may have arisen in part from the fact that a seismic array in Norway detected a magnitude 2.5 event in the Novaya Zemlya region on January 13, 1996.  A seismic signal of this magnitude would correspond to a well coupled underground explosion of a few tons of TNT, or about a thousand-ton decoupled explosion.  Later data analysis by the independent Incorporated Research Institutions for Seismology determined, however, that the event was an earthquake -- not at the test site, but under the sea."

Suzanne L. Jones and Frank N. Von Hippel, "Transparency Measures for Subcritical Experiments Under the CTBT," Science and Global Security, 1997, Vol.6, p.291, 292-3.

Verification of a zero-yield test ban is further complicated by above-ground, contained subcritical testing.  All kinds of subcritical tests raise concerns that there may be a small, difficult to detect nuclear yield, which can expand the range of experiments and the information that can be gained from them.  Contained subcritical tests exacerbate this problem, because there may be little external evidence (which could be obtained, for example, by satellite surveillance) that the tests are being conducted.  Although in principle contained tests might allow a higher degree of transparency, no regime for international observation or inspection of subcritical tests is in place.