Jeffrey LewisSubcritical Testing at Lop Nor

I am off to Cambridge for less than 24 hours to participate in a workshop entitled, “Getting to Zero: US and Chinese Perspectives on Near Term Cooperative Steps to a World Free of Nuclear Weapons.”

I’ve been spending a lot of time thinking about China’s efforts in terms of stockpile stewardship, remanufacture, modernization, what-have-you.

The United States and Russia are the only two countries that admit to conducting so-called “subcritical” nuclear tests — tests that “involve chemical high explosives and fissile materials in configurations and quantities such that no self-sustaining nuclear fission chain reaction can result. In these experiments, the chemical high explosives are used to generate high pressures that are applied to the fissile materials.”

China probably does, too.

In 1999, the State Department claimed that “China has initiated … a program at its Lop Nur test site” to evaluate the safety and reliability of its nuclear weapons. The National Academies suggested that these activities would include “hydrodynamic tests with flash radiography and subcritical tests.”

Press reports have suggested ongoing activity at the Chinese test site, but I only recently noticed the best evidence of Chinese subcritical nuclear testing at Lop Nor hiding in plain sight.

I was reading Mark Fisk’s Accurate Locations of Nuclear Explosions at the Lop Nor Test Site Using Alignment of Seismograms and IKONOS Satellite Imagery (Bulletin of the Seismological Society of America 92:8, pp. 2911–2925, December 2002). It’s not as boring as it sounds, really. Really. Anyway, I noticed this image:

Pretty. Each circle identifies an adit — an opening into a tunnel where China conducted nuclear tests during the 1990s. There are four — labeled A, B, C and D. Fisk explains:

Table 2 lists the approximate locations of the tunnel adits. More than one adit may actually exist within the circles of Figure 4. Adit A, with the clearest evidence of excavation activity, appears to be the newest, followed by adit B, then adit D. Adit C, with the most eroded features, appears to be the oldest. This suggests that either adit A or B was used for the nuclear test on 29 July 1996. Similarly,
significant effects of erosion near adit C suggest that this tunnel was likely used for the earliest Chinese tunnel shots in this zone, dating back to 17 October 1976. There is no clear evidence from the image of which adit was used for the nuclear test on 25 September 1992, although adits B and D appear to be reasonable candidates.

Now go look at the site in GoogleEarth today (well, in 2005) :

There are now five adit areas. To the west of Adit Alpha, a new area has appeared. Here is a close up:

The adit on the left was clearly constructed after Fisk’s image was taken in 2000.

The construction is, I believe, strong evidence that China — like Russia and the United States — conducts subcritical nuclear tests.

Now this shouldn’t surprise (or alarm) anyone — subcritical tests are not prohibited by the Comprehensive Test Ban Treaty or any other arms control treaty. The US has even made available video of one of its subcritical tests.

But the new adit may shed light on a minor controversy that erupted in 2001, coincidental to the EP-3E spy plane incident. In April 2001, the United States detected test preparations at Lop Nor. It seems that a debate erupted within the intelligence community about whether the preparations were for low-yield nuclear tests or merely subcritical tests. Someone leaked all of this to Bill Gertz.

The Chinese apparently conducted three tests in June and July 2001, followed by a fourth in December 2001. Seismic stations did not detect any seismic activity. US intelligence “sniffer” aircraft did not get a whiff of any gas that would indicate a chain-reaction. This is persuasive, though not conclusive, evidence that the testing was subcritical.

The subcritical tests may have occurred in the tunnel leading from the new adit. This would neatly explain why the intelligence community thought it might be a yield test — the Chinese dug a new hole next to the old ones used for yield testing. You don’t need a vivid imagination or raging xenophobia to be a little suspicious.

Of course, it is also perfectly reasonable to conduct a subcritical test at a test site. That is what the United States and Russia do. If you do it right, there shouldn’t be any risk. But, a declassified 1995 intelligence report stated that “a hydronuclear experiment went out of control at China’s Southwest Institute of Fluid Physics in 1993, resulting in contamination that forced scientists to seal off a portion of the laboratory” and “may have motivated the Chinese to transfer their hydronuclear activities to Lop Nur.” Good call, boys.

This is just conjecture and circumstantial evidence — the new adit may be post-2001. Gertz reported that the Chinese conducted a subcritical experiment in June 1999, before the new adit area was constructed, suggesting they probably either branched off an older tunnel or conducted the test someplace else at Lop Nor. (Or Gertz is wrong.)

Maybe this is just a giant, underground dodgeball arena. But I suspect not. This seems worth following up, starting with pinning down the construction timeline for the new adit.


  1. peter Zimmerman (History)

    Just a minor matter: by the convention of the CTBT negotiators in the US, “tests” involve nuclear yield and are prohibited. “Experiments” have no perceptible nuclear yield (i.e. remain subcritical) and are permitted.

    The demarcation, as far as close-up observers are concerned, between sub-critical and critical is pretty bright. It is difficult, especially if you haven’t conducted a lot of near-critical safety tests (in the US that means trying to keep nuclear yield below four pounds TNT equivalent) to go just barely critical and not go significantly critical. Yield goes up a lot faster than linearly with the compression beyond criticality. So I can believe that the Chinese made a costly mistake in a hydronuclear experiment, especially if it was designed to get close to critical or exceed the the threshold by a small amount.

  2. Wm. Robert Johnston (History)

    Whether the UK has admitted to subcritical testing, the US has admitted it on their behalf. The DOE has announced that two subcritical tests at NTS were joint US/UK tests, Vito in 2002 and Krakatau in 2006:
    Krakatau Subcritical Experiment Conducted