CCNet DEBATES, 28 April 1998


Six weeks after 1997 XF11 made the world's news headlines, the
NEO-search community has come to a general conclusion: There is no
threat whatsoever from this PHA [that is, excluding any 'effects that
would be highly unusual']. Yet some questions remain. One of the most
controversial is still unresolved: at what stage of observation and
orbit calculation could this certainty have been achieved?

Early last week, Alan W Harris (JPL/NASA) notified me about IAUC 6879.
He interpreted Brian Marsden's circular as an admission that as early
as 21 December 1997, two weeks after Jim Scotti's discovery of XF11,
'a collision could have been ruled out.' Since I came to a different
reading of IAUC 6879, I asked Alan to clarify a number of questions
which had been raised by the ongoing debate.

I would like to thank Alan for trying to explain some of the underlying
problems with asteroid 1997 XF11. I would show-off if I were to claim
that I now understand all (or even most) of the problems involved. But
at least I have the feeling that I am getting closer to comprehend what
the actual mathematical complexities (and uncertanties) are.

As far as I am concerned, the whole episode has been, more than
anything else, a tremendous learning experience, not just for me but, I
guess, for most list members of the CCNet. We shouldn't underestimate
this positive side-effect, in spite of all the controversy. As a
result, perhaps twice or three times as many people may be able to
participate in, assess or simply understand the accuracy of complex
orbit calculations next time around - at least that's what I hope will
be the case in the future.

Benny J Peiser



On Tue, 21 April Benny J Peiser [BJP] wrote to Alan W Harris:

Dear Alan

I am relieved to see the first attempts of reconciliation within the
NEO-search community. I very much hope that this positive direction and
a cooled atmosphere will make it easier to critically assess what the
REAL problems with XF11 have been and what we can learn from this
experience. I also hope that despite all the controversy, no bridges
have been burnt so that we can continue to cooperate in the full
knowledge that disagreement may remain on some issues.

Since our controversy has taken place under the eyes of a critical
public (which is, I believe, crucial for upholding public confidence in
NEO research), it would certainly be helpful to convey to the
interested public the preliminary results of this re-assessment within
the NEO community. I will ask Brian whether the MPC would have any
objections to make those parts of the IAUC 6879 which deal with 1997
XF11 available for the CCNet [see annotated IAUC 6879, CC DIGEST

Yet some questions remain which I find difficult to understand.

1.) On Mon, 20 April you wrote [about IAUC 6879]:

> Although it is a bit concealed (MPEC 1997-Y11 contains only
> observations from the date of discovery, Dec. 6, to Dec. 21, a 15-day
> arc!), Marsden has accepted the fact that already by then a collision
> could be ruled out. 

What you are saying is that after as little as 15 days of observation
of 1997 XF11 "a collision could [have been] ruled out". I understand
that as late as 10 days ago, one of your colleagues has calculated - on
the basis of the 1997/98 data - a hypothetical collision of XF11 with
earth. Tell me if I got this wrong, but how - on this basis - can one
rule out any impact hazard?

On Sun, 26 April, Alan W Harris [AWH] wrote:

I believe you have got it a bit wrong. One way to test the robustness
of a conclusion (especially a negative one) is to force a positive
solution and see what contradictions arise.  In the 1997 XF11 case, one
can insert a false "observation" into the data set, essentially an
"observation" of XF11 in October 2028 that corresponds to an impact on
the Earth, and then compute the orbit, to see what happens to the fit
to the real observations. This is a quite common and straightforward
way to demonstrate in easy to understand terms whether a conclusion is
firm or not. 

The statement that the Earth lies 30-sigmas outside of the error
envelope, and hence the formal collision probability is something like
10^(-300), loses any possible understanding. So, we can, figuratively
speaking, grab hold of the orbit, and "pull it over" to pass through
the Earth in 2028, and see what is the very least displacement from the
observations in 1997-8 that can allow that to be true. 

There are some practical difficulties in carrying out this computation,
for example some programs may have difficulty when the computed orbit
passes through a perturbing body, and fitting an "observation" where
the solution distance tends to zero (the false observation on the
Earth's surface) can be difficult. Anyway, several colleagues have
pursued this method and have produced orbits to evaluate the
plausibility of a collision trajectory based on various subsets of the
1997 XF11 observations.

But here's the rub:  this "solution" is a mathematical solution to a
hypothetical situation posed; it is not necessarily a physically
allowed "solution" to a real situation, indeed it is demonstrably NOT
an allowed solution, which is exactly the reason for examining it. Let
me describe in general terms what has been found from these orbits.

The solution through all of the data, including 1990, left "observed
minus computed" residuals to the actual observations as large as 10
arc-minutes, or 1/3 the diameter of the moon.  It is inconceivable that
any modern competent observer could make such horrendous errors in
observation. Even Tycho Brahe did far better without a telescope, and
with only the star catalogs of his day. 

Furthermore, the fit residuals to this forced solution are not random
-- they fall along a very narrow curve when plotted with time,
indicating that each of the many individual observers would have to be
making the same gross errors as all the others, of many arc-MINUTES, to
within a fraction of an arc-SECOND. Clearly this is inconceivable, and
demonstrates that the collision trajectory is impossible, in much more
graphic and humanly understandable terms than stating a probability of
"10^(-300)", or the blunt, "that's zero, folks."

This is not quite the end of the story. Everyone agrees that, with the
inclusion of the 1990 observations, an impact is totally out of the
question. Since the largest residuals occur for the 1990 observations,
the above solution tends to overwhelm what might be done to better fit
the 1997-8 observations. Ted Bowell, at some sacrifice to his normal
time last night for a beer, made an attempt at the problem using his
orbit determination program, inserting a couple of synthetic
"observations" a couple hours before the "impact" in 2028.  He was able
to force a new solution through these points and preserve the impacting
trajectory, but avoid the problems that earlier caused his program to
blow up.  He thus employed this technique to recompute an orbit,
excluding the 1990 observations, which would impact the Earth in 2028,
and computed the residuals to the 1997-8 observations.  The pattern is
the same as for the orbit including the 1990 observations, although the
amplitudes of the systematic deviations are much smaller.  In
Declination, the residuals range from +10 arcseconds at the time of
discovery to about -6 arcseconds on March 3-4. In right ascension, the
residuals run from -4 arcseconds at the time of discovery to +22
arcseconds on March 3-4. As before, the plot of residuals vs. time
fall along systematic trends to within an arcsecond or two.  Thus we
are left with the same conclusion: it is really inconceivable that
these observations, coming from diverse observatories with different
instruments could be that wrong, and yet so systematically concordant
with each other but not with the forced "solution."  To be sure, the
last word is not in, and work is still in progress fine-tuning this
result.  However, the basic conclusion that an impacting trajectory
could be ruled out appears robust, and I don't think it is a fair
assessment to imply that it has taken more than a month to certify that
conclusion as "correct."

Finally, you question whether this situation is true for such a short
arc of data as 15 days. Here I am merely quoting IAUC 6879. Brian said
it, I didn't. Paul Chodas has run calculations in terms of his
"10^(-large number)" collision probabilities that would tend to support
that conclusion. In this case, the statement in IAUC 6879 actually went
further than I expected to see.


There are other questions which have been raised by the events of the
last month:

Why did various NEO researchers - even after 88 days of observation -
come up with significantly different miss distances? How can these
calculations differ so greatly if they use exactly the same data?


This is another excellent question, but again may take a bit of space
to explain thoroughly. Recall my analogy of a couple weeks ago to train
tracks. The key point is that all of the various solutions you mention
fall "on the tracks", and therefore are concordant with each other,
even though numerically (in one dimension) they sound significantly
different. I'll draw a little picture below, which is only intended to
be schematic, not quantitative.

                     1                  2               3

                                  X                 _
                                                   |_| Earth

In the above sketch, the parallel lines represent the "error ellipse"
that I have likened to a pair of railroad tracks.  For the Chodas
solution, the "ellipse" extends out a few meters (around 10 feet) in
both directions from your computer screen. The error ellipse defines
the range of space within which the real solution (the actual
encounter, in this case) might occur. Any solution which falls
essentially in the center of this range of uncertainty can be
considered "concordant" with any other solution that also does, even if
they are not precisely coincident. 

Thus solutions 1, 2, and 3 are concordant with one another, especially
when we remember that the error envelope extends many meters out of the
range plotted.  But the solution X is not concordant, even though it
would appear to predict a miss distance in the same range as 1, 2, or
3.  The fact that the various orbit computers obtained solutions like
1, 2, and 3 which are mutually concordant even if a bit different gives
us confidence that all are essentially correct. There were no solutions
like "X".

You may reasonably ask why there should be any difference at all among
solutions using exactly the same data. One reason is that the problem 
is quite complex, and is of a class known as "non-linear," so that the
solution is obtained by successive approximations. To solve such a 
problem, one makes an initial guess, then computes the "covariance
matrix" which describes how the errors in fit to the observations vary
with each (and all together) of the solution parameters, in this case
the orbital elements. This covariance matrix allows one to derive a
next level approximation to the solution parameters that will better
fit the data. But because the problem is non-linear (in a sense, you
are taking a straight line step down a curved path), you don't quite
end up at the exact solution, just a better approximation of it. 

This  same covariance matrix allows you to estimate when a solution is
"good  enough," that is, when the difference of the present solution
from the optimum one is so small that the residual difference has no
physical significance or predictive value. In our example above,
solutions 1, 2, and 3 are all physically equivalent; one is no more
likely than the other to be a better estimate of what nature will
actually do. So one possible explanation of why the solutions are
different may be that the different computers simply chose different
thresholds of "good enough" in deciding to quite making successively
finer improvements in their orbits.

Different choices of initial orbits can lead to different
"trajectories" of convergence on the final solution, so that it is not
surprising that various computations end up at slightly different spots
in the error envelope. As long as a solution is sensibly in the center
of the envelope, then they are all equally valid and concordant. In
physical terms, for the example above, it is impossible to predict from the
data available whether the true answer will be 1, 2, 3, or something else
within the error envelope, so all can be taken as equally valid.


How much observational data (days, weeks, months, years?) is required
to calculate a NEO-orbit with such accuracy that all  estimations do no
longer differ significantly?


This is bound to vary from case to case. At the outset, I would like to
emphasize that the early solutions for XF11 did not "differ
significantly" from one another, which is the point of my answer to the
previous question. As to when an orbit solution is good enough to rule
out a collision, that is a more difficult question. The easiest case is
when the Minimum Orbit Intersection Distance (MOID) can be shown to be
greater than some distance such that we can be sure that a collision is
impossible, no matter where in their respective orbits the Earth and
the asteroid are.

To make another analogy, this is like verifying that two different
aircraft are flying at different elevations such that no collision is
possible. This turned out to be the case for 1997 XF11, but only
barely. This is why we could rule out a collision, but still not define
very precisely just how close the encounter would be. Imagine two
aircraft flying in crossing paths but 1000 feet different in elevation.
they might come as close as 1000 feet to each other, but then again,
they might not get closer than miles apart, depending on where along 
their courses they are with respect to the other.

But suppose the orbits are intersecting, or in the above analogy, the
aircraft are flying at the same altitude. Then it becomes a much more
time consuming task to rule out a collision. 1997 XF11 provides a good
lesson in how things might proceed, if we "tickle" the orbit a bit in
the computer (or our mind) so that it is on a collision course. The
first test is the MOID: if you're not on the tracks, you're not going
to get hit. In the case of 97 XF11, we could determine reliably that we
were "not on the tracks" from only the 1997-8 data. 

The next question is, if you are "on the tracks," are you within the
range of uncertainty along the orbit? From the 1997-8 data alone, the
answer was, yes, we are in the range of uncertainty along the track, so
if the solution had put the Earth "on the tracks," the 1997-8 data
would not have been sufficient to rule out a collision. It would still
be improbable, because the range of uncertainty was several hundred
times the width of the Earth. When we added in the 1990 data, the
uncertainty range along the track shrank to a much smaller distance,
and in particular, the Earth was no longer included within that range.
That is, the Earth not only was laterally displaced from the error
ellipse, it was far off from the end of the ellipse, like so:

                                                    error ellipse


But it may not have been so. Suppose it had turned out not only that
the Earth was "on the track", but also "within the range" of the new
error ellipse using the 1990 data. Note that the new error ellipse is
still longer than the Earth, so if we were within that box, we would
still not know whether we are going to be hit or not. I believe I have
drawn it correctly; the error ellipse, last time I checked, still has a
long dimension longer than the Earth, so if we were in it, we would
need to improve the orbit further in order to rule out (or in) an

In about two years, in 2000, 1997 XF11 will be easily observable for
most of the year, but far away from the Earth. It is possible that
observations at that time would suffice to shrink the error ellipse so
that its longest dimension is less than the size of the Earth. I tend
to doubt it, or if so only barely. In 2002, the asteroid will pass
within radar range, and if radar observations are obtained, the error
ellipse can be expected to shrink by two or three orders of magnitude.
Following that, we would be able to say with quite absolute certainty
whether or not an impact would occur. So, for 1997 XF11, which I
believe is a fairly typical example, if it had been on a collision
trajectory, it probably would be until 2002 when we get radar
observations before we could say for sure that it would (or would not)
impact.  But that is still 26 years warning. LET ME EMPHASIZE THAT THIS
NEAR ON A COLLISION TRAJECTORY.. We are neither "on the track" nor
"within the range" of a collision path.

The good news is, the above scenario is extremely unlikely, and I can
show simply why that is so. There are a thousand or so "XF11's" out
there to be found. Using XF11 as an example, after a month or two of
tracking, we can project ahead for a century whether the orbit
intersects the Earth's (MOID less than the Earth's dimension), and we
can define the position in the orbit to within a few million km. Of the
100 or so already known, none have a MOID smaller than the Earth's
dimension, although a couple (like XF11) come close enough that one
might need to become concerned about specific close encounters, that
is, detailed location in the orbit. Although the statistics of this are
poor, we can estimate that of the thousand or so left to discover, only
of the order of 10 will be found to have MOIDs close enough to zero
that we need to consider specific encounter circumstances, that is,
position in orbit.

But that position is known to within a couple million km (for a century
or so), and a typical NEA orbit track is several billion km long
(circumference, or whatever that is for an ellipse). So the chance that
there will be an encounter with the Earth such that the MOID point in
the orbit is contained within the error ellipse is only around one in
1000, per Earth orbit, for a given asteroid, or only around 10% in
a century. For all the asteroids (about 10) for which the MOID is low
enough to matter, that's still only a chance of about one asteroid.
This is a very rough argument, and because the assumptions used are
very generous, the bottom line is that it is unlikely (not highly
unlikely, but somewhat unlikely) that even one asteroid which will be
discovered by a Spaceguard Survey will be found in an orbit which
cannot be quickly resolved, within a month or two, to be harmless for
at least a century. This is a fundamentally different perspective than
has been repeated in much of the impact hazard literature, and one
which should affect how we approach the question of checking and
announcing a claim of a hazardous encounter.


How big has to be the miss distance in order to "rule out" collision?
How big, to make it highly unlikely? How big, to make it impossible to
rule it out?


The short answer here is, one Earth radius from the center of the
Earth, if you are sure of the trajectory. We have targeted spacecraft
to swing by the Earth only a few hundred km up, with no ill effects on
either the Earth or the spacecraft. But I am being a bit facitious. 
What counts, of course, is the uncertainty. 

As I have indicated above several places, the uncertainty in the "cross
track" direction, as we say in spacecraft navigation, is typically a
thousand km or so for a near-Earth asteroid. So if that track (cf. the
first schematic diagram above) misses the Earth by a few thousand km
(well, maybe 10,000), we can be comfortable enough. However the
along-track uncertainty can easily be a million km. So if we had a case
with a MOID of zero, we would do well to keep a very close eye on the
object even if the nominal miss distance in some future encounter is
estimated to be a million km or more. Thus you really can't state a
minimum or maximum distance for "worry" vs. "no worry,"  because the
uncertainly envelope is so terribly elongated.


What errors and inaccuracies can effect the estimation of
orbit-calculations and how should they be taken into account?


A typical error source is catalog errors in the reference stars used to
derive positions from the astrometric observations. For this reason, we
often see errors of as much as an arcsecond, even when the measurement
precision is a small fraction of that. Now and then gross errors occur,
such as misidentifying an image (of a reference star or the asteroid).
For that reason one needs more than just a few observations to be sure
everything is OK. The theoretical minimum needed to compute an orbit is
three positions. It's pretty dangerous to do anything with less than
about 10, so that any wrong measurements will jump out. 100 is better.

Obviously, the more observations you have from the more independent 
observers, the better your chances are that there are no obvious
errors, or that if there are they will jump out at you and can be
eliminated. The things to check in a solution are whether there are any
outliers in the distribution of residuals (individual bad 
observations), if there systematic trends in the residuals (indicating
unmodelled effects or an inaccurate solution), or a non-Gaussian
distribution of residuals (suggesting a bad fit or faulty solution),
and finally, the "devil's advocate" test I described in answer to your
first question, where you force the trajectory to do what you are
testing against, namely a collision, and see what that does to the fit
to the observations.


In view of the fact that there is no general agreement about the actual
impact probability (due to the biases in the impact crater record and
the asteroidal/cometary flux), why does background probability play
such an important role in the assessment of the impact


The background probability is important only to as an
order-of-magnitude gauge of importance of individual events. I think
your statement that there is "no general agreement" is a bit strong. I
think within an order of magnitude, there is pretty general agreement. 
If you don't like "background level," I could couch the same concept a
bit differently. I think we can agree upon the a priori impact
probability of a single object, given only that it exists and its orbit
crosses the Earth's (that is, perihelion inside of, and aphelion
outside of, the Earth's distance). The instant an object is discovered,
that may be all we know, and the intrinsic hazard from that one object
can only be stated as being at that level, which happens to be about
10^-8 per year, since the random collision lifetime of a single object
is about 10^8 years (a bit less, actually). As the orbit becomes better
known, what we are learning is where the object will go for the next
century or so, and as an obvious corollary, where it WON'T go. That is,
certain bits of space have an increased probability of being
intersected, and other bits of space must therefore have a decreased
probability of being intersected (the sum total is constant of course,
the asteroid actually will be one place or another, all of the time).

You can think of the space we occupy in terms of a probability
function. If we happen to be sitting on a peak of that function, we
have a better (or worse?) than random chance of being hit; if in a
valley, a lower than average probability. After a very short amount of
observation of a given object, that probability function already
becomes a very peaked affair, with most of the "chance" concentrated
along a very narrow track, and the rest of the space left far below
"background" level. So in almost every case, we can expect that as soon
as we start to gather enough observations to get even a rough
preliminary orbit, the probability of an impact will go down, not up.
It is only in a very rare case that we should see an increasing
probability at all. This is the same point I was making a couple
questions above. 

To return to the point of your question, I would say that a well
authenticated impact probability of even 10^-6, which is way below the
"background level" in 30 years, would suffice to raise my eyebrows, but
the chance that any one object out there would have such a high
probability, after even a month or two of observations, is very small.


If 1997 XF11 would have been on a potential collision course, would
that have altered your view that impact would be highly unlikely due to
the background impact probability? In other words, what is more
important for any future risk assessment - orbit calculations or impact
probability statistics?


I'm not sure I understand the point of your question. If one finds that
something is going to happen, then it's going to happen, and the
probability of its occurrence becomes moot. However, we must be mindful
of the admonition, "extraordinary claims require extraordinary proof." 

This brings me back a third time to the point that it is intrinsically
unlikely that even one object will be found for which an impact cannot
be ruled out from a properly analyzed set of observations of only a
couple months duration. Thus a claim of such a hazard, based on such a
set of observations, must be regarded as "extraordinary," and treated
accordingly. That doesn't mean disbelieved out of hand, but it does
mean it demands careful scrutiny and checking.


These are just some of the many questions which have been occupying my
mind for the last four or five weeks. Perhaps you are happy to answer a
couple of them so that we can continue our debate on a more advanced


Whew! Done! Sorry to have carried on so long. I think this has done me 
good in terms of clarifying some of the points in my own mind (you
learn more from teaching than you do from studenting, it's well known).
I hope the above dissertation has similarly clarified some points for

BJP: Thanks for your time and effort.


CCNet DIGEST, 28 April 1998

    Duncan Steel <>

    Phil Burns <>

    Phil Burns <>

    Phil Burns <>

    Jane E Allen, AP

    Phil Burns <>

    H. Dypvik & R.E. Ferrell, University of Oslo


From Duncan Steel <>

Dear Benny,

With regard to the interesting item Bob Kobres sent you, given a
LEGEND, here is an addendum.

There is another legend retold in the mythology of the Paakantji people
of the Darling River (from around Wilcannia in western New South
Wales). This tells of a foreseen falling star which brought fire and a
following flood, killing many people and leaving behind strange stones.
Sound familiar?  The story is told in the (picture) book 'The Story of
the Falling Star' by Elsie Jones, Aboriginal Studies Press, Canberra,
1989 (ISBN 0-85575-199-1).  A speach bubble from a lady's picture on
the back cover (perhaps Elsie Jones) says "This story is so old we
don't even know how old it is...Malkarra was a special kind of
person...He told the Paakantji people something bad was going to
happen...They didn't trust him...If only they had listened to Malkarra
they would've been gone when the star fell..."  Various places in the
book feature drawings of an incoming bolide which would not be out of
place in the forthcoming Hollywood movies.

This legend and book is mentioned as a postscript to:

D. Steel & P. Snow, 'The Tapanui region of New Zealand: A 'Tunguska' of
800 years ago?' pp.569-572 in Asteroids, Comets, Meteors 1991 (eds. A.
Harris & E. Bowell), Lunar and Planetary Institute, Houston, Texas,
U.S.A. (1992).

...which is largely concerned with legends of the Maoris of New Zealand
which, it was suggested, might involve a Tunguska-type event. That
paper was featured in a half-page article in New Scientist (5 October
1991, page 19).

Returning to the content of the newspaper article sent by Bob Kobres,
it is well-recognized that the east coast of Australia shows signs of
one or more large tsunamis during the Holocene.  I am not sure
how/whether the dating is done. The major sand dune formations are
aligned so as to be side-on to the north-east (towards Hawaii, say) and
ocean-bottom boulders have been found over 40 metres above sea level (I
am informed; I am no geomorphologist & I have not gone looking myself
to verify these things).

Duncan Steel


From Phil Burns <>

SKY & TELESCOPE magazine now has a web site about impact events
to complement the cover story in the June 1998 issue:

The site includes an article by Gerrit L. Verschuur and an annotated list of
web pages about the impact threat by Stuart J. Goldman.

-- Phil "Pib" Burns
   Northwestern University, Evanston, IL.  USA


From Phil Burns <>

Two movies with plotlines about large impact events are scheduled for
release soon here in the U.S.  Their release will probably result in
a new round of questions about impact events from the general public.

DEEP IMPACT from Paramount is written by Michael Tolkin and Bruce
Joel Rubin  and stars Robert Duvall, Tea Leoni, Elijah Wood, Vanessa
Redgrave, Maximilian Schell, Leelee Sobieski, and Morgan Freeman. 
Steven Spielberg is one of the executive producers. This movie has
its own web site at:

describing the story. The Discovery Channel here in the States will
air a special entitled DEEP IMPACT NIGHT on May 4, 1998. I assume 
this is some kind of tie-in with the movie.

ARMAGEDDON stars Bruce Willis. This movie also appears to have its
own web site:

I have not been able to access this web site successfully so I don't
have any more information about the movie.

-- Phil "Pib" Burns

From Phil Burns <>

Several folks on the Cambridge list have been warning folks for some
time of the dangers to satellites posed by a possible Leonid meteor
storm next November. The mainstream press here in the U. S. has
finally started paying attention. Jane E. Allen of The Associated
Press offers a report dated April 27 entitled "Meteoroids Threaten
Satellites."  You can read this at:

The report stems from the Leonid Meteoroid Storm and Satellite Threat

-- Phil "Pib" Burns



MANHATTAN BEACH, Calif. (AP) - In November, the Earth's atmosphere will
be hit with the most severe meteor shower in 33 years, a bombardment of
debris that could damage or destroy some of the nearly 500 satellites
that provide worldwide communications, navigation and weather-watching.

The debris consists only of particles - some thinner than a hair and
most no larger than a grain of sand - but they are hurtling through
space so fast that they can have the destructive power of a .22-caliber

As a result, about 200 commercial and military satellite operators,
insurers and scientists began brainstorming here Monday about what they
can do to prepare, such as turn off spacecraft or turn them away from
the stream of particles. The two-day gathering is called the Leonid
Meteoroid Storm and Satellite Threat Conference.

"The consequences are still virtually unknown. There has not been a
meteor storm since the onset of the modern space age. Nobody planned
for it," said Peter Brown, a physics and astronomy graduate student at
the University of Western Ontario who advises satellite operators.

The particles, known as meteoroids, are vastly smaller than the
asteroids that could one day slam into Earth, and none are expected to
come anywhere near the surface of the planet when they strike this
November and again in November 1999.

But before the particles burn up in Earth's atmosphere, they could poke
holes in solar panels, pit lenses, blast reflective coating off
mirrors, short out electronics with a burst of electromagnetic energy,
even reprogram computers, said Edward Tagliaferri, a consultant to the
Aerospace Corp., a nonprofit organization.

In 1993, for example, a meteor struck the European Space Agency's
Olympus satellite and destroyed its directional control, rendering it

"What if you get unlucky?" Delbert Smith, a Washington lawyer who
represents international networks and satellite operators, asked at the
conference. "Who's going to explain to the major corporations your
satellites aren't there anymore?"

While only a couple of satellites might get disabled - and some cost as
much as $500 million - all of them will suffer surface damage, said
David Lynch, a scientist with the Aerospace Corp.

Military satellites are better shielded because most are built to
withstand nuclear assault. But unlike commercial spacecraft that can be
turned off temporarily, military satellites "can't afford to be off the
air," Tagliaferri said.

The Hubble Space Telescope - which suffered minor surface damage in the
1993 shower - will move to protect itself against Leonid damage by
turning away from the stream of particles, an option being considered
by many satellite owners.

First reported by Chinese astronomers back in 902, the Leonid meteoroid
storms - so-named because they are found in front of the constellation
Leo - become intense every 33 years. They occur when Earth passes
through a trail of dust left behind by the comet Tempel-Tuttle.

Scientists aren't sure when the heaviest showers will occur - Nov. 17,
1998, or Nov. 18, 1999.

The spectacular showers will be visible this year across the Western
Pacific and Eastern Asia; the 1999 showers will be visible in the
Middle East, Eastern Europe and Central Asia. Storms last 90 minutes to
two hours.

Back in 1966, when fewer than 100 satellites circled the Earth, the
comet produced peak showers of 144,000 meteors each hour and no major
damage. This year, with more than five times the number of circling
spacecraft, some experts think the rate could be 5,000 to 100,000 an

But astronomer Donald Yeomans of NASA's Jet Propulsion Laboratory in
Pasadena put the rate as low as 500 to 2,000 particles per hour. And
Brown agreed that the rate won't be as high as it was in 1966.

(C) 1998 The Associated Press.


From Phil Burns <>

The Columbus (Ohio) Dispatch for April 26, 1998 contains a report by
David Lore reporting on the controversy surrounding the origin of
Serpent Mound in Adams County, Ohio (USA).  A volcanic origin for this
feature has been the orthodox viewpoint, although a few suggested an
impact origin. Geologist Mark Baranoski now suggests that there is
"unequivocal evidence" for an impact origin..  His conclusions are
disputed by other geologists who have examined Serpent Mound.

The full story may be found at

-- Phil "Pib" Burns
   Northwestern University, Evanston, IL.  USA


H. Dypvik*) & R.E. Ferrell: Clay mineral alteration associated with a
meteorite impact in the marine environment (Barents Sea). CLAY
MINERALS, 1998, Vol.33, No.1, pp.51-64


More than 50 samples from a Barents Sea borehole near the Mjolnir
Structure (an extraterrestrial impact feature) were used to investigate
changes in the clay assemblage associated with the submarine impact.
Seismic evidence, the presence of shocked quartz and a prominent Ir
anomaly restricted the potential impact affected zone to a 10 m
interval, straddling the Jurassic/Cretaceous boundary. Increased
abundance (up to 30 wt%) of a smectite, a randomly interstratified
smectite-illite with 85% smectite layers, forms the basis for a
two-layer oceanic impact clay model that differs from published
terrestrial cases. The smectite is assumed to represent
seawater-altered impact glass from the ejecta blanket material that was
mixed with resuspended shelf sediments by the collision generated
waves. The smectite-rich interval is almost 5 m thick. It is overlain
by a coarser unit (similar to 2 m thick) containing abundant smectite,
shocked quartz grains, and anomalous Ir contents at its base. The
smectite-rich interval may have originated as a density/turbidity
current, generated by the impact and the collapse and erosion of the
crater rim. Seawater alteration of volcanic glass and changes in the
tectonic regime of the provenance area, or changing oceanic current
circulation patterns could produce similar variations in the clay
mineral assemblage. The most compelling evidence for the possible
impact derivation of this clay assemblage is the direct association
with the Mjolnir Impact Structure and associated mineralogical and
geochemical anomalies. Copyright 1998, Institute for Scientific
Information Inc.

The CCNet is a scholarly electronic network. To subscribe, please
contact the moderator Benny J Peiser at <>.
Information circulated on this network is for scholarly and
educational use only. The attached information may not be copied or
reproduced for any other purposes without prior permission of the
copyright holders.

CCCMENU CCC for 1998