Last changed 1 September 2001
On 22 September 1979 around 00:53 GMT, the Vela 6911 satellite detected the characteristic double flash of an atmospheric nuclear explosion apparently over the Indian Ocean or South Atlantic. The test location was later localized at 47 deg. S, 40 deg. E in the Indian Ocean, in the vicinity of South Africa's Prince Edward Island, by hydroacoustic data. Due to the position ambiguity of the initial detection (the Vela optical sensors were not imaging sensors and could did not detect location), the location is variously described as being in the Indian Ocean or South Atlantic. The characteristics of the light curve indicated that it was a low kiloton explosion (approximately 3 kt). The hydroacoustic signal indicated a low altitude explosion. A major and lingering controversy erupted over the interpretation of this apparent detection.
The Vela satellite program was an nuclear detonation (NUDET) detection system setup after the 1963 limited test ban and was designed to detect nuclear explosions in space and (later) air. There were two groups of Vela satellites developed. The original Vela were equipped only with sensors for space detection and were launched in three pairs between 1963 and 1965. They operated for at least five years, far beyond their nominal design life of six months. A second generation called Advanced Vela were launched in 1967, 1969 and 1970. These satellites added "bahngmeters" - optical sensors for detecting atmospheric tests - and had a nominal design life of 18 months, but were later rated with a seven year lifespan, although they were all operated for more than ten years, with the last one being turned off in 1984 -- after 14 years of successful operation [JPL 2001]; [Astronautix 2001].
Vela 6911 is presumably one of the Advanced Vela pair launch launched on 23 May 1969 (perigee 77,081 km, apogee 145,637 km, inclination 61.6 deg), and had thus been operating over ten years at the time of the 1979 detection.
The Vela satellite system had previously made 41 similar detections of atmospheric tests, each of which had been subsequently confirmed through other means. The detection came at a bad time for the Carter administration which would be under pressure to take definite action if the detection were accepted as accurate. Inescapably it seemed that either Israel, South Africa, or both, would be implicated. Consequently a panel of scientists from academia known as the Ruina Panel, after its head Dr. Jack Ruina, was created to review the reliability of the Vela data. Since this satellite was operating past its expected lifespan, and its electromagnetic pulse (EMP) sensor was inoperative, questions about the reliability of the detection were raised. The panel ultimately concluded in a report released in the summer of 1980 that the signal "was probably not from a nuclear explosion. Although we cannot rule out that this signal was of nuclear origin".
This conclusion has cast a pall over public confidence in the ability of the U.S. to unambiguously detect clandestine nuclear explosions for over twenty years.
Advanced Vela 5
The instruments used by the Vela satellites for detecting atmospheric nuclear explosions are called "bhangmeters". These are optical sensors that record light fluctuations on a sub-millisecond time scale. All atmospheric nuclear explosions produce a unique and easy to detect signature: an extremely short and intense flash, followed by a second much more prolonged and less intense emission of light. The initial flash is typically 1 millisecond long, and although it emits only about 1% of the total thermal energy of the fireball, it is actually the point of maximum brightness for the fireball. The second peak may take from hundreds of milliseconds to several seconds to develop, depending on the size of the explosion, and lasts a comparable period of time.
This phenomenon occurs because the surface of the early fireball is quickly overtaken by the expanding hydrodynamic shock wave. This shock wave acts as an optical shutter, hiding the small but extremely hot and bright early fireball behind an opaque ionized shock front which is comparatively quite dim.
No natural phenomenon is known that can imitate this signature. In fact it is reported that no false alarms have ever been detected with a Vela bhangmeter. Every other double-flash detection has later been confirmed to be an actual nuclear test.
According to Seymour Hersh, the idea of referring this detection to an advisory panel was floated before any potential problems with the detection had been noted. An urgent meeting to discuss the handling of this event was held in the White House situation room soon after the intelligence report on the incident reached the Oval Office. Among those attending were National Security Advisor Zbigniew Brzezinski, his aide for global issues Gerald Oplinger, deputy director of the Arms Control and Disarmament Agency Spurgeon Keeny, and Presidential Science Advisor Frank Press. At this meeting the probability of a nuclear test was placed at 90 percent or better. Either Keeny or Press (accounts of the participants vary) suggested convening a panel - at least as much as a delaying tactic as an effort to ensure that the data was carefully evaluated [Hersh 1991].
There is no question that a confirmed detection of a nuclear test would have put the Carter administration in a very difficult position. President Carter had placed great emphasis on nuclear non-proliferation. The administration had been a troubled one, with the recent collapse of a major ally in the Middle East (Iran) as one of many problems. The upcoming re-election campaign was certain to be an uphill battle. If Israel were to be linked to a nuclear test (as seemed likely, it it was real) the political damage from imposing sanctions, or not imposing sanctions, would likely be severe.
Within several weeks the eminent membership of the panel had been selected - Jack P. Ruina, professor of electrical engineering at MIT and an alumnus of several defense think tanks, was the titular head, of the eight other members Nobel laureate Luis Alvarez, Wolfgang Panofsky of Stanford, and Richard Garwin of the Thomas Watson Research Center at IBM were the key players.
The administration succeeded in keeping the detection secret until 25 October 1979, when ABC television reporter John Scali broadcast the story after having been briefed by contacts at the Pentagon.
From the outset the panel was given guidelines tailored to help give the Carter administration cover - they were tasked to investigate whether the detection had been a false alarm including the possibility that it "was of natural origin, possibly resulting from the conincidence of two or more natural phenomena...". Given this mandate and focus it was perhaps inevitable that the result of the panel's work would be the most plausible possible way to explain away the detection. It should be remembered that at the time of the panel first convened, no reason to doubt the detection had been identified.
Problems were found with the Vela satellite data though - the two bhangmeter readings did not agree on the flashes brightness, perhaps because the aged sensors were no longer equally sensitive. This discrepancy, and the lack of confirming data from the inoperable EMP sensor, emerged as the chief reasons for casting doubt on whether a nuclear test had actually occurred. Discrepancies had been observed in Vela signals from previous confirmed atmospheric nuclear tests however [LANL Daily News Bulletin 1997].
During the panel's months of deliberations, concluded in July 1980, a variety of pieces of corroborating evidence surfaced.P>One of the clearest indications was from ocean acoustic waves detected by hydrophones. The hydrophone data indicated signals both from a direct path originating near Prince Edward Island and from a reflection of Scotia Ridge in the Antarctic and the Antarctic ice shelf. Analyses of these signals conducted by the Naval Research Laboratory (NRL) confirmed that they had been generated at a time and location consistent with the Vela 6911 detection and that their intensity was consistent with a small nuclear explosion on, or slightly under, the ocean's surface. This evidence alone, if accepted as valid, should be sufficient to confirm the accuracy of the detection [LANL Daily News Bulletin 1997]; [Albright 1994b].
The radio telescope at Arecibo, Puerto Rico detected an anomalous traveling ionospheric disturbance (that is, an upper atmosphere wave) moving southeast to northwest during the early morning hours of 22 September 1979, something researchers had never before witnessed. Powerful explosions create ionospheric disturbances from the direct transmission of the upward propagating shock wave to the ionosphere. But Los Alamos space scientist Lew Duncan, one of the researchers who originally connected the ionospheric disturbance to the event that night, said in 1994 that he was still not fully convinced that what the dish at Arecibo detected was a nuclear test [Albright 1994b].
Frank Barnaby cites additional ionospheric data collected later by NRL, and confirmation from Air Force early warning radar signals made on 22 September, that was not considered by the Ruina Panel though he does not describe the evidence in greater detail [Barnaby 1989; pg. 17-18].
Unfortunately efforts to detect the one completely unambiguous "smoking gun" signature of a nuclear explosion, guaranteed to silence skeptics, namely radioactive fallout, failed. The U.S. government quickly launched a major effort to collect samples of the fallout cloud, but Air Force attempts to sample the fallout failed to enter the low-pressure air mass that had been over the detonation site at the time of the explosion (what the problem was I do not know). However Dr. Van Middlesworth detected low levels of iodine-131, a short-lived radioactive fission product, in sheep thyroids in the states of Victoria and Tasmania in western Australia soon after the event. Studies of wind patterns confirmed that fall-out from an explosion in the southern Inidian Ocean could have been carried there [Barnaby 1989; pg. 17].
An summary of the panel's deliberations and conclusions is provided by one its most distinguished members - the Nobel winning physicist Luis Alvarez:
[Alvarez 1987, Chapt. 14]
The two ways to read the one good satellite record from which the explosive yield was determined didn't agree as well as usual. Drawing on my bubble chamber experience, I asked to see a selection of the satellites' "zoo-ons", events so strange they belonged in a zoo. This idea was new to the DIA, but since their records were stored on computer tape they needed only a week to put their zoo together. Rich [Richard Muller], Dick [Richard Garwin], and I found a steady degradation in record quality among these zoo-ons from confirmed explosions to events at which no one would look twice. Although the event we were studying had some of the characteristics of a nuclear explosion, only one of the two satellite sensors recorded it. Moreover, there was no indication from earlier or later records that the sensor that failed to record the event was malfunctioning. Both sensors looked at a large area of the Earth's surface, so it was hard to believe that one sensor could see a nuclear blast and the other could not.
Someone on the committee proposed that a micrometeorite might have struck the satellite and dislodged a piece of it skin. Reflecting sunlight into the optical system of one sensor but not into that of its neighbor, the debris might have caused the questionable event. We constructed a believable scenario based on the known frequency of such micrometeorite impacts that reproduced the observed light intensity and pattern.
I doubt that any responsible person now believes that a nuclear explosion occurred because no one has broken security, among South Africans or elsewhere. U.S. experience teaches that secrets of such import can't be kept long. After the United States tested its first megaton-scale thermonuclear weapon, which completely evaporated the small Pacific island of Elugelab, stories about a disappearing island reached U.S. newspapers as soon as the task force steamed into Pearl Harbor and sailors had time to call home.
Many people think that solving a scientific puzzle is an exercise in logic that could be carried out equally well by a computer. To the contrary, a scientific detective's main stock-in-trade is his ability to decide which evidence to ignore. In out DIA briefings we were shown, and quickly discarded, confirming evidence from a wild assemblage of sensors: radioactive Australian sheep thyroids, radiotelescopic ionospheric wind analyses, recording from the Navy's sonic submarine-detection arrays that supposedly precisely located the blast from patterns of sound reflected from bays and promontories on the coast of Antarctica.
Inevitably, then, I had a real sense of deja vu when the House subcommittee's [on the Kennedy-assassination police-radio recordings] acoustics experts pinpointed the location of the open mike by triangulating the reflections of supposed gunshot sounds from the building in Dealey Plaza. The National Academy of Sciences committee showed conclusively that the motorcycle with the open mike wasn't even in Dealey Plaza at the time the tape was recorded.
Alvarez's discovery that Vela detections were part of a continuum of detections of variable quality, and that this detection was less clear cut than others, provided a defensible rationale for dismissing it as a real detection. It showed that there existed a class of ambiguous detections, that could include false events, and perhaps real ones as well. Any event in the lower range of "good" detections could be treated as suspect.
Alvarez's account here that only one sensor detected the test is at variance with other accounts ([LANL Daily News Bulletin 1997]; [Albright 1994b]) that state a difference in the recorded intensities, but not complete non-detection. This is an important point, because a complete failure by one sensor would seriously weaken the case of a nuclear test, while similar signals from both sensors (even though they differed in strength) would make the micrometeroid theory much more difficult to credit. It is perhaps relevant to point out that Alvarez is mistaken in his implication that the House subcommittee hearings were held after the Ruina Panel, when in fact they were held several years before. This raises questions about how reliable his memory of this affair is overall.
A feature that stands out from Alvarez's account is the apparent summary dismissal of all corroborating evidence, one which he offers no real rationale for (other than to suggest that this is how real science its done). Although Alvarez proposed a statistical model (the micrometeoroid theory) to explain away the detection, he does not seem to consider another type of statistical model that tends to support the possibility of a nuclear test. This is a causal model in which a hypothesized cause is evaluated by considering the likelihood of each of a set of ambiguous evidences is a consequence of it, none of which is a "smoking gun". Each piece of evidence may have an alternate explanation, or a background rate of occurrence, unconnected with the the hypothesized cause, but taken together the hypothesis may be a far more likely explanation for the whole set, than assuming that each is an independent and uncorrelated red herring. Dismissing each piece one at a time, as the panel seems to have done, is a suspect procedure.
Another argument that Alvarez makes - that the absence of decisive revelations (known to him) by 1987 was conclusive evidence against its occurrence - is clearly flawed.
First, not all mysteries are ever resolved or resolved quickly. The disappearance of Judge Crater, to pick one hoary example, was never explained. The circumstances surrounding the disappearance of the last Romanovs remained unexplained for 75 years; the disappearance of Raoul Wallenberg for 50. There are specific reasons that make these things possible - only limited groups of people knew the truth, and they had a variety of strong motivations not to reveal what they knew, and perhaps serious obstacles to doing it. In particular, Alvarez's experience with the Ivy Mike thermonuclear test was hardly similar to the situation in Israel, or in South Africa at the time. For starters - no story could ever have reached any Israeli newspaper, for the entire Israeli press is under military censorship which bans any reference to the Israeli nuclear program. The United States has never had any similar censorship regime, even at the height of involvement in world wars. Further, Israel has demonstrated willingness to resort to kidnap operations on foreign soil, secret arrests and incarceration, and intimidation to maintain secrecy, and perhaps would not cavil at resorting to assassination. In the 70s and early 80s South Africa was also a repressive nation with de facto censorship, and extra-legal hit teams to enforce social order.
Second, if the existence of stories supporting the hypothesized test are taken as a requisite for believing that it occurred then this condition was well satisfied by 1991 when Hersh recounted detailed stories about an Israeli-South African test collaboration in The Samson Option [Hersh 1991; pg. 271-272]. In his treatment Hersh takes the reality of the test as a given, a matter beyond dispute. He refers to former Israeli government officials who indicated that the flash was a test of a low yield nuclear artillery shell, and was actually the third such test in the area. At least two Israeli navy ships had sailed to the site and a contingent of Israelis, and the South Africa navy was observing the test. The tests were conducted under cover of bad weather, but a gap in the clouds allowed the detection.
Similarly Barnaby claims that South African naval ships were operating in the area the night of the test, citing an African Educational Fund study on the incident [Barnaby 1989; pg. 17].
Hersh reports interviewing several members of the Nuclear Intelligence Panel (NIP), which had conducted their own investigation of the event. Those interviewed included its leader Donald M. Kerr, Jr. and eminent nuclear weapons program veteran Harold M. Agnew. The NIP members concluded unanimously that it was a definite nuclear test. Another member - Louis H. Roddis, Jr. - concluded that "the South African-Israeli test had taken place on a barge, or on one of the islands in the South Indian Ocean archipelago" [Hersh 1991; pg. 280-281]. He also cite internal CIA estimates made in 1979 and 1980 which concluded that it had been a test.
The U.S. Naval Research Laboratory conducted a comprehensive analysis, including the hydroacoustic data, and issued a 300-page report concluding that there had been a nuclear event near Prince Edward Island or Antarctica [Albright 1994b].
The failure of the panel to affirm the existence of the test seems to reflect the nature of the panel and its mandate more than a failure of the evidence. Dave Simons of Nonproliferation and Arms Control Research and Development (NIS-RD) said that similar discrepancies had been observed in Vela signals from earlier confirmed atmospheric tests and is quoted as saying: "The whole federal laboratory community came to the conclusion that the data indicated a bomb," [LANL Daily News Bulletin 1997].
The principle question that has hung in the air for the last 20 plus years has been more along the lines of "whose test was it?" than "was it really a test?". The choices were basically:
The possible theories of responsibility remain the same as in 1979, although more information is now available to flesh them out.
By the time of the Vela detection it was universally believed that Israel had a sophisticated nuclear weapons program (this was well before the Vanunu revelations in 1986). South Africa was known to be pursuing a weapons program, but the status of their effort was unknown to the outside world, although the US had detected preparations for a nuclear test site in the Kalahari desert in 1977, [Burrows and Windrem, 1994]).
During the years following the Vela detection unsourced reports periodically surfaced ascribing the test to one of the above possibilities.
As described above, in 1991 Seymour Hersh in The Samson Option quoted a number of source Israeli sources as saying that the test was a joint Israeli and South African operation [Hersh 1991]. If it was an Israeli test, one must speculate exactly how it was done. Hersh claims Israeli navy ships were sent - but Israeli has a small blue water Navy most of which is based in the Mediterranean. It would require a very unusual extended deployment out of the area to reach the Indian Ocean for a test - and the whereabouts of Israel's navy ships around the time of the test would provide a means for evaluating its plausibility. I do not know whether anyone has publicly documented information about this - although I have been told by someone convincingly representing himself as a former South African naval intelligence officer that in fact Israel's ships were not unaccountably absent around this time. Even if so, the possibilities exist of staging the test using commercial ships - operated by Israeli crews possibly under a front company, as an all-airborne operation using air-to-air refueled transports, or by using South African platforms.
Barnaby reports an account from African Educational Fund study that U.S. reconaissance planes operating in the area around the time of the test were intercepted by South African military aircraft and forced them to land. Also cited is circumstantial evidence like a literature search done by the U.S. National Technical Information Service (NTIS) at the request of a South African representative "on nuclear explosions and the seismic detection of nuclear explosions, including the flight plans, predicted orbit plans and operations of the Vela satellite", the only such request ever received [Barnaby 1989; pg. 18-19].
On the other hand Waldo Stumpf, of the Atomic Energy Corporation of South Africa, argues that South Africa did not have the means to conduct a test at that time, but also states that "South Africa was certainly not responsible and was also not involved with anybody else, in this incident [Stumpf 1995].
Prime Minister F.W. de Klerk announced in March 1993 that South Africa had built nuclear weapons, and since that time additional information has periodically come forth. It was revealed that SA had indeed developed and manufactured nuclear weapons (gun-type devices using highly enriched uranium) but no tests (beyond a single zero-yield lab test) were disclosed. The information that was made available tended to disconfirm the hypothesis that South Africa conducted the test. The IAEA has apparently been able to confirm that whatever discrepancies exist between South Africa's HEU inventory and its production records (and some are inevitable) the amount is too small to hide the HEU required for a test. Further, South Africa's accounts of its weapon development activities indicate that its first device was not complete until months after the incident. From documents made available to it, the IAEA believes that the first nuclear device was not manufactured until November. This first device was an experimental one named "Melba" which was said to be kept for research and demonstration purposes throughout the program (which ended in 1989).
In 1994 another claim surfaced:
Contradicting these statements is Commodore Dieter Gerhardt, a convicted Soviet spy who was at the time the commander of the Simonstown naval base near Cape Town. After this release from prison, Gerhardt settled in Switzerland. In Februrary 1994, he told Des Blow of the Johannesburg City Press that the flash was produced by an Israeli-South African test code-named "Operation Phenix". Gerhardt, who said he was not yet ready to reveal the full facts, stated that although he was not directly involved in planning or carrying out the operation, he had learned of it unofficially.
Gerhardt was quoted in the February 20, 1994 City Press: "The explosion was clean and was not supposed to be detected. But they were not as smart as they thought, and the weather changed - so the Americans were able to pick it up."
Gerhardt told me in a March interview that no South African ships were involved in the event. He declined to provide any more details."
[Albright 1994b; pg. 42].
It is impossible to assess whether Dieter Gerhard's account has any basis in fact. Some parts of his statements are interesting. The assertion that "The explosion was clean ..." suggests that it was a neutron bomb (which has been alleged before about this event) which is the only kind of low yield device to have reduced fallout. His statement that weather pattern changes caused its detection is interesting, since conclusive detection by this means has never been made public - the Australian fallout report (and a New Zealand one before it) were both subject to dispute. By his own admission of course, he had no direct connection with the project.
Quite a stir erupted in 1997 when in a 20 April 1997 article that appeared in the Israeli daily newspaper Ha'aretz, South African Deputy Foreign Minister Aziz Pahad was quoted as confirming that the 22 September 1979 flash over the Indian Ocean was indeed from a South African nuclear test. The article said that Israel helped South Africa develop its bomb designs in return for 550 tons of raw uranium and other assistance.
Initially this seemed to conclusively decide the nature of the Vela incident, at least as far as its participants went (the possibility of undisclosed Israeli participation remained). Pahad's office later responded that his remarks were taken out of context. His press secretary told the Albuquerque Journal in an article dated 11 July that Pahad had said only that there was a "strong rumor" that a test had taken place, and that it should be investigated. In other words - Pahad was not commenting on actual knowledge of a test, but was repeating rumors that had been circulating for many years.
Alvarez's point about the absence of evidence being evidence of absence in the case of South Africa today holds much more weight than it did in 1987. South Africa has not only revealed its formerly secret nuclear arsenal and weapons program, but has dismantled both, and the apartheid regime that built them is no more. In the era after the Truth and Reconciliation Commission has unearthed many hidden and unpleasant truths regarding the National Party government (and crimes authored by the same), it is increasingly difficult to believe that all participants in nuclear testing, or those with authoritative knowledge of it, would all still remain silent.
If the test was a South African nuclear device then at least some of the information earlier released by the South African government would have to have been falsified. In particular, the information provided to the IAEA that South Africa did not construct its first nuclear explosive device until November 1979, two months after the mysterious flash, and that the first batch of highly enriched uranium was kept in an experimental device until 1989. IAEA investigations of detailed Valindaba production records, and inventories of its highly enriched uranium, appear to support the claim that any inventory irregularities could not hide enough HEU to fashion a bomb.
Whether the device was South African or Israeli in origin, Gerhard's account (if true) indicates joint Israeli-South African participation in the test. South Africa has now admitted direct Israeli involvement with the South African weapons program, at least to the extent of providing weapon design advice and exchanging material support. Israel had previously been known to provide certain special materials - in particular rather large amounts of tritium - to the weapons program. But nothing has come from the South African government indicating direct Israeli involvement in the testing of a nuclear device, and certainly not the lead role implied by Gerhard's assertion that 'no South African ships were involved'.
NB: An LASL (now LANL) study of physical phenomena possibly related to the Vela detection is available on-line in Acrobat (.pdf) format. This document appears to have been prepared prior to the availability of the Arecibo data mentioned earlier. A description and abstract of the document is given below. The downloaded file is 1.4 megabytes.
NTIS No: LA-8672/HDM Title: Evaluation of Some Geophysical Events on 22 September 1979 Author(s): Hones, Jr., E. W. ; Baker, D. N. ; Feldman, W. C. Performing Organization: Los Alamos Scientific Lab., NM. Sponsoring Organization: Department of Energy, Washington, DC. Contract No: W-7405-ENG-36 Date: Apr 81 Pages: 20p Abstract: TIROS-N plasma data and related geophysical data measured on 22 September 1979 were analyzed to determine whether the electron precipitation event detected by TIROS-N at 00:54:49 universal time could have been related to a surface nuclear burst (SNB). The occurrence of such a burst was inferred from light signals detected by two Vela bhangmeters approx. 2 min before the TIROS-N event. The precipitation was found to be unusually large but not unique. It probably resulted from passage of TIROS-N through The precipitating electrons above a pre- existing auroral arc that may have brightened to an unusually high intensity from natural causes approx. 3 min before the Vela signals. On the othe hand, no data were found that were inconsistent with the SNB interpretation of the 22 September Vela observations. In fact, a patch of auroral light that suddenly appeared in the sky near Syowa Base, Antarctica a few seconds after the Vela event can be interpreted (though not uniquely) as a consequence of the electromagnetic pulse of an SNB.