Minor planet 4179 Toutatis is an Apollo type object with a very small
orbit inclination (i = 0.degrees 47), hence it has a possibility to ap
proach closely the Earth tan encounter to within 0.01 a.u. is expected
in 2004) and might be a good candidate for a future collision with th
e Earth. We collected 640 astrometric observations of Toutatis from th
e period 1934-1997 to improve the orbit. We had to include a nongravit
ational term into equations of motion expressed by a secular change (a
) over dot of the semimajor axis a of the Toutatis orbit to obtain a f
ully satisfactory solution of the orbit determination. A value (a) ove
r dot = -0.16 x 10(-10) is two orders smaller than that determined in
the case of short-period comets with known nongravitational effects. T
o investigate the long-term motion of Toutatis we numerically integrat
ed the equations of motion by recurrent power series taking into accou
nt perturbations caused by the eight planets from Mercury to Neptun, t
reating the Earth and Moon as separate bodies, and also by the four bi
ggest asteroids. We randomly varied the orbital elements to examine th
e Toutatis' motion for a number of different orbits. We present a new
method of the random orbit selection which allows us to find a set of
different orbits but representing well all the observations used for t
he: orbit correction. Our results confirm a conclusion found by other
authors that Toutatis orbit is exceptionally chaotic. Therefore, we ar
e not able to predict the motion of Toutatis further than for 300 year
s. However, our integrations spanning 1500 years showed that the evolu
tion of position of the descending node of Toutatis' orbit might go al
so in such a direction that the orbits of Toutatis and of the Earth wo
uld intersect in the future. Hence a possibility of the Toutatis-Earth
collision is not excluded but it is completely unpredictable. To inve
stigate conditions of a hypothetical collision of a minor planet with
the Earth we made the following numerical simulation. Based on the Tou
tatis' orbit we deduced such orbital elements for a fictitious minor p
lanet ''Fatum'' that a shape of the orbit was very similar to that of
Toutatis, but we knew in advance that ''Fatum'' would certainly collid
e with the Earth in September 2004 and we calculated values of the imp
act parameters. We created a set of 638 artificial observations of ''F
atum'' in 1988-1997 for the same dates and with the same random observ
ational errors like those of Toutatis. Then we corrected the ''Fatum's
'' orbit for different observational intervals to examine the exactnes
s of the impact prediction in 2004. We found that in 1993 we would be
sure that the collision is inevitable, and in 1997 we could determine
an impact area on the Earth's surface in range of a square of 100 x 10
0 km. We show that if we knew the impact date so early we could undert
ake an action to avoid the collision by trying to change the ''Fatum's
'' heliocentric velocity only by one cm/sec.