The Envelope of Ballistic Trajectories and Elliptic Orbits

Eugene Butikov
Saint Petersburg State University, Saint Petersburg, Russia

Abstract
Simple geometric derivations are given for the shape of the “safety domain” boundary for the family of
Keplerian orbits of equal energy in a central gravitational field and for projectile trajectories in a uniform
field. Examples of practical uses of the envelope of the family of orbits are discussed and illustrated by
computer simulations. This material is appropriate for physics teachers and undergraduate students studying
classical mechanics and orbital motions.

Key words: safety domain, Keplerian orbits, family of orbits, foci of the ellipses, envelope of trajectories,
minimal shooting speed.

1 Introduction: The Envelope of Projectile Trajectories

A famous old problem of classical mechanics concerns projectiles launched over the Earth from a given
point with a given velocity whose direction is varied. All the projectiles start from the origin in various
directions and move under the uniform field of gravity in the absence of the air resistance. This results in
a family of parabolic trajectories (see Fig. 1) whose envelope defines the border of the so-called “safe do-
main.” It is well known (since the work of Torricelli1 back in 1644) that the envelope of these trajectories is
a parabola (or a paraboloid of rotation about the vertical axis in three dimensions). This boundary separates
the points in space that can be reached by a projectile from those that are out of reach from the given initial
point for the given initial speed.

Figure 1: Family of parabolic trajectories traced by projectiles launched from one point in various directions
with equal initial speeds, and their envelope (dashed line).

Many different ways for deriving the equation of the envelope are discussed in the literature. It is
scarcely possible to reference here all relevant papers. In particular, the problem of the safe domain in a
uniform field is treated in Ref. [1], where successive launches with varied initial angle are considered, as
well as simultaneous firing in all directions. With some recent articles and books [2]–[5] it is possible to
trace back many previous contributions to the issue.
1 An italian physicist and mathematician Evangelista Torricelli (1608–1647), best known for his invention of the barometer, studied

a lot about projectiles. “Perhaps his most notable achievement in the field of projectiles was to establish for the first time the idea of
an envelope: projectiles sent out at the same point and the same speed in all directions trace out parabolas which are all tangent to a
common paraboloid. This envelope became known as the parabola di sicurezza.” Philip J. Robinson, The Mathematical Gazette, Vol.
78, No. 481 (Mar., 1994), pp. 37–47.

1
 However, these trajectories are parabolas only approximately, when the initial speed is sufficiently small,
and when they are confined to a region within which the gravitational field is nearly uniform (the approxi-
mation of the “flat Earth”). Actually, the projectiles move along portions of ellipses whose remote common
focus is located at the Earth’s center. The greater the initial speed, the greater the deviations of real trajec-
tories from parabolas. Hence we can ask a natural question: what is the real (exact) shape of the envelope
of these trajectories?
More generally, let us imagine a rocket, launched from the Earth, rising vertically, and at the highest
point of its flight exploding into many fragments (point masses) that fly off in all directions with equal
speeds relative to the non-rotating geocentric frame of reference. In the absence of air resistance, the
further motion of the fragments occurs solely under the action of Earth’s gravity, whose magnitude falls off
as the inverse square of the distance from the center of the Earth. If the initial speed is large enough, some
of the fragments become satellites, orbiting along various elliptic Keplerian paths. If some of the ellipses
cross the surface of the Earth, these trajectories are portions of such ellipses.
If the initial speed of the fragments is small, a portion of such an ellipse above the ground is approxi-
mately a parabola. This parabolic shape we usually assign to the trajectory of a projectile in the approxima-
tion of the “flat Earth" (that is, in the nearly uniform gravitational field) and in the absence of air resistance.
In any case, all the trajectories are confined to a particular spatial region, if the initial speed of the
fragments is smaller than the escape velocity. The boundary of this region is an axially symmetric surface
whose axis of symmetry passes through the center of the Earth and the point at which the explosion occurs.
What is this surface? What is the shape of the envelope of all elliptic orbits of the family? This question
may be of interest to physics teachers and undergraduate students studying classical mechanics and orbital
motion.

2 The Family of Elliptic Orbits

Several orbits of such families and their envelopes are shown in Fig. 2. These figures and those that follow
were generated with the help of one of the interactive simulation programs of the software package [6]
developed by the author. We note that the program does not “know” anything about Kepler’s laws or
analytical solutions; it simply integrates numerically the differential equations of motion under the central
gravitational force of the Earth. For the family of orbits in Fig. 2(a), the initial speed v0 of the satellites
is smaller in magnitude than the “circular velocity” vcirc (the velocity of a satellite in a circular orbit that
passes through the initial position), while in Fig. 2(b), v0 is greater than vcirc .
One focus, at the center of the Earth, is common to all elliptic orbits. The second focus, for the family of
orbits traced by satellites launched from a given point with a given initial speed, always lies on a given circle
whose center is located at the common starting point. These circles are shown by dashed lines in Fig. 2.
The radius of this circle equals the distance between the initial position and the highest point reached by the
fragment that is fired vertically upward from the initial position. These results will be proved in the next
section.
Since the speeds of the fragments are equal, and since the motion of each fragment begins at the same
spatial point, the total energies (per unit mass) of each are equal. The total energy depends uniquely on the
major axis of the elliptic orbit in the inverse-square central field (see, for example, Refs. [7]–[9]). Therefore,
the major axes of all the orbits are also equal, and, according to Kepler’s third law, the periods of revolution
of all fragments are equal. That is, all of the fragments whose orbits do not intersect the Earth’s surface
return to the initial position simultaneously.
A detailed analysis (see the next section) shows that the boundary of the safe domain is a surface of
revolution of an ellipse (spheroid) whose foci are at the center of the Earth and at the initial position. The
dimensions and eccentricity of the spheroid are determined by the position of the initial point and by the
initial speed of the fragments. We will give a rigorous proof and a derivation of the parameters of this
bounding surface, relying on the geometrical properties of Keplerian orbits.
For the limiting case of small initial speed v0 of the fragments (much smaller than the circular velocity
vcirc for the starting point), this bounding surface shrinks into a very narrow (degenerate) ellipsoid spanned
across the foci located at the initial point and the center of the Earth. The portion of this ellipsoid near
the starting point can be approximated by the familiar envelope of the parabolic trajectories in the uniform
gravitational field.

2
 a b

Figure 2: Families of elliptic orbits traced by fragments launched from one point over the Earth in various
directions with equal speeds, and the envelope of orbits. Thin straight lines show major axes of the ellipses;
dashed circle centered at the initial point is the locus of their second foci. (a) – initial velocity v0 =
0.75 vcirc , (b) – v0 = 1.20 vcirc .

3 The Shape of the Envelope

Below we give a simple geometrical proof of the fact that the bounding surface for the family of elliptic
orbits with equal energies and a common initial point is a spheroid, generated by the rotation of an ellipse
about the axis of symmetry ZN (see Fig. 3). One of the foci of this ellipse is located at the center of the
Earth E, and the other at the initial point S.
The fragment whose initial velocity (velocity at point S in Fig. 3) is directed upward rises directly to the
highest point Z located at a distance rmax from the Earth’s center. Then it falls toward the Earth along the
same line. The trajectory of this fragment is a portion of the rectilinear segment joining the highest point
Z with the center of the Earth E. We can consider this segment as the limiting case of an infinitely narrow
ellipse spanned across points Z and E. The foci of this degenerate ellipse lie at the ends Z and E of the
segment.
Clearly the highest point Z lies on the bounding surface. The distance rmax = rZ of this point from
the Earth’s center can be easily calculated by equating the total energy of the fragment at this point Z to the
total energy at the initial point S, located at the distance r0 from the center of force:

v02 GM GM
− =− . (1)
2 r0 rmax
Here G is the gravitational constant and M is the mass of the Earth. It is convenient to express the gravita-
tional parameter GM of the Earth √ in Eq. (1) in√ terms of the escape velocity vesc for the initial point S (recall
that the escape velocity vesc = 2GM/r0 = 2vcirc is the minimal initial speed that must be imparted to
a projectile in order for it to reach infinity):
( )
1 1 v02 r0
= 1− 2 ; rmax = rZ = . (2)
rmax r0 vesc 1 − (v0 /vesc )2
√
If the initial velocity equals the circular velocity for the initial point, that is, if v0 = vcirc = GM/r0 ,
then Eq. (2) gives rmax = rZ = 2r0 : The distance of the highest point Z from the Earth’s center is twice
the distance r0 of the initial point. The distance rmax = rZ tends to infinity if v0 → vesc .
We can easily find one more point on the desired boundary, namely, the point N on the opposite side
(see Fig. 3). It coincides with the apogee (or with the perigee if v0 < vcirc ) of the elliptic orbit of the
fragment whose initial velocity at S is directed horizontally (is directed transversely to the radius vector).

3
 Z

v0
S F1
E

F2

F3

Figure 3: Locus of the second foci of the elliptic orbits (the dashed circle).

The distance rN of this point from the Earth’s center can be calculated with the help of the laws of energy
conservation and angular momentum conservation (r0 v0 = rA vA ):
r0
rN = . (3)
(vesc /v0 )2 − 1

Next we find the locus of the foci of all the orbits of the fragments. All orbits have a common focus at
the center of the Earth (Fig. 3), and so the locus of this set is the point E. The locus of the set of second
foci (points F1 , F2 , F3 , and Z in Fig. 3) is a circle whose center is located at the initial point S, and whose
radius is equal to the distance |SZ|, measured from S to the most remote point Z (the dashed circle in
Fig. 3).
Indeed, for any orbit of the family, the sum of two distances of each point on the orbit from the foci of
this orbit equals the major axis of the orbit. The major axes of all the orbits of equal energy constituting
the family, as we already mentioned, are equal to one another. Their lengths are equal to the length rmax of
the segment EZ. We can consider this segment to be the major axis of the degenerate elliptic orbit of the
fragment whose initial velocity is directed upwards. All orbits of the family pass through the initial point
S, and the distance SE from this common point to the focus E for all the orbits is the same and equals r0 .
Consequently, the distance between S and the second focus F1 , F2 , or F3 also must be equal for all the
orbits. Hence the second foci F1 , F2 , F3 of all orbits of the family lie on the circle whose center is at the
initial point S and whose radius equals |SZ| (see the dashed circle in Fig. 3).
Next we prove that the curve whose rotation generates the boundary is an ellipse. The ends of the major
axis of this ellipse are located at Z and N , and its foci are located at the initial point S and the Earth’s center
E. The shape of the boundary is derived below from the well-known geometric properties of the elliptic
trajectories.
We start by considering the following auxiliary construction (Fig. 4): We draw one more circle whose
center is at the Earth’s center E and whose radius equals rmax = |EZ|. This circle passes through Z, which
lies on the bounding surface.
Now let us consider the problem of finding the orbit of the family that passes through an arbitrary target
point T that lies within the second circle just drawn. Choosing T as a center, we draw a third circle tangent
to the second circle at point B, as shown in Fig. 4. The second focus of the desired orbit passing through T
must lie on this circle, because the sum of two distances from the foci again must be equal to rmax . And at
the same time this second focus of the orbit must lie on the first circle (whose center is at the initial point

4
 a Z b Z

v0 v0
2
S S
1 F2
E E
T F
F1 B

T
B
N N

Figure 4: Geometrical proof of elliptical shape of the bounding surface (see text for details).

S and whose radius equals |SZ|). This circle is shown by the dashed line in Fig. 4. We examine three
possibilities:

1. The third circle (whose center is at the target point T ) intersects the first circle (the locus of second
foci of the orbits) at two points (F1 and F2 in Fig. 4(a)). Then there exist two orbits of the family that
pass through the given target point T . These orbits are labeled by numbers 1 and 2 in Fig. 4(a). The
second foci of these two orbits lie at the two points of intersection (F1 and F2 ).
2. The third circle has no common points with the first circle. Then no orbit of the family passes through
T . If follows that the chosen point T lies outside the bounding surface.
3. Lastly, the third circle grazes the first circle, thus having a single common point F with it (see
Fig. 4(b)). Then only one orbit of the family passes through the target point T . In this case point T
must lie on the bounding surface. At point T this single orbit grazes the envelope surface. The foci
of this orbit are located at F and the center of the Earth E.

We can see from Fig. 4(b) that in the third (grazing) case, the sum of distances from the target point T
to the center of the Earth E and to the starting point S equals the radius rmax of the second circle plus the
radius |SZ| of the first circle. This sum is independent of the position of point T on the boundary. That is,
the sum has equal values for all points of the boundary. Since points E and S are fixed, and since the sum
of their distances from T is the same for all T , we have proved that the locus of the boundary points for
the region occupied by the orbits of the family is an ellipse whose foci are at the center of the Earth E and
starting point S.
We note that the exploitation of geometrical properties of ellipses allowed us to easily find the envelope
of the family of orbits without tedious calculations.
The eccentricity of the bounding ellipse can be found as the ratio of the distance r0 between its foci to
the major axis rZ + rN (see Fig. 4(b)). Using Eqs. (2) and (3), we find

r0 v 2 − v02
e= = esc
2 + v2
. (4)
rZ + rN vesc 0

For a target lying on the bounding surface (see Fig. 4(b)) there exists a unique trajectory of the family
of elliptic orbits with a given energy. This trajectory grazes the boundary just at the target point T . To hit

5
the target, the initial velocity v0 of the projectile must have a definite direction (along the bisectrix of the
angle T SZ in Fig. 4(b)).
Indeed, the trajectory of the projectile fired to the given target T with the minimal starting speed is a
portion of the ellipse passing through S and T . One focus of this Keplerian ellipse is located at the center
of the Earth, while the second focus belongs to the circle whose center is at the starting point S and whose
radius equals |SZ| (see Fig. 4(b)). Therefore this second focus is located at point F , at which the segment
ST joining the starting point with the target intersects the mentioned circle. Knowing locations of both foci
for the elliptic trajectory of the projectile, we can easily find the firing angle with the help of the optical
property of the ellipse. Since a light ray emitted from the focus E to S must be reflected at S by an elliptical
mirror towards the second focus F , the tangent to the ellipse at S (and hence the direction of initial velocity
v0 min ) is the bisectrix of the angle F SZ (or T SZ, which is the same; see Fig. 4(b)).
Any target lying within the bounding surface can be reached by two trajectories of the given family. One
of them (the trajectory labeled 1 in Fig. 4(a)) first passes through the target point T , and only then grazes
the boundary. The other (labeled 2 in Fig. 4(a)) is generated by a projectile launched in a different direction,
which is closer to the local vertical line at the starting point. This trajectory first grazes the boundary, and
only then passes through the target point.
If we increase the initial speed of the fragments, the bounding surface expands and its shape becomes
more spherical. Indeed, as we can see from Eq. (4), the eccentricity e of the bounding ellipse becomes
smaller and tends to zero as the initial speed approaches the escape velocity. If the initial speed v0 equals
the circular velocity vcirc for the initial point, the distance between the foci of the bounding ellipse is one
third its major axis. That is, the eccentricity of the ellipse is 1/3 if v0 = vcirc .
As the initial speed tends to zero, the eccentricity of the envelope approaches unity. Then the apexes
(the ends of the major axis) of the bounding ellipse approach its foci, and the ellipse becomes very narrow
and stretched, being spanned over the initial point S and the center of the Earth E. This limiting case of
a degenerate ellipse corresponds to the parabolic shape of the above-mentioned bounding surface for the
trajectories of the fragments moving within a restricted spatial region (in the vicinity of the initial point S)
in which the gravitational field can be regarded as uniform.

4 Useful Applications of the Envelope

Knowing the boundary surface can be useful in solving various problems concerning the orbital motion.
For example, we can easily find the minimal firing speed of a projectile needed to hit a given target from a
given starting point. Suppose we have a target T (see Fig. 4(b)) at a given location. Let the target location be
determined by distance rT = |ET | from the force center E (the center of the Earth) and distance lT = |ST |
from the given starting point S (whose distance from the force center is r0 = |ES|). What is the minimal
initial speed and what should be the firing angle?
The firing speed is minimal if the target T lies on the bounding surface. Since this boundary is an
ellipse, the sum of distances |ET | and |ST | from T to its foci (points E and S) is equal to the major axis
of the bounding ellipse: |ET | + |ST | = rZ + rN . The sum |ET | + |ST | is just the sum of given distances
rT and lT to the target from E and S. Let us denote this sum as b: rT + lT = b. Thus, we can equate
this given value b to the major axis rZ + rN , which has already been calculated above, when we derived
expression (4) for the eccentricity of the bounding ellipse:
2
vesc + v02
b = rZ + rN = r0 . (5)
vesc − v02
2

Solving this equation, Eq. (5), for v0 , we obtain the desired minimal firing speed:

b − r0
v02 min = vesc
2
. (6)
b + r0
Equation (6) shows that for a given position of the starting point S the minimal firing speed depends
only on b, that is, on the sum of distances rT and lT that determine the target location (b = rT + lT ).
According to Eq. (6), the firing speed is zero if b = r0 , that is, for any target that lies on the segment SE

6
 √ point and the center of the Earth. The minimal firing speed v0 min tends to the escape
joining the starting
velocity vesc = 2gR2 /r0 as the target is moved away to infinity (as b → ∞).
As we have shown above, the firing angle can be easily found with the help of the optical property of
the ellipse. Since a light ray emitted from the focus E to S must be reflected at S by an elliptical mirror
towards the second focus F , the tangent to the ellipse at S (and hence the direction of initial velocity v0 min )
is the bisectrix of the angle F SZ or T SZ (see Fig. 4(b)).

5 Shooting Equatorial Targets from the Pole

Next we discuss particular examples of trajectories traced by the projectiles launched with minimal initial
speeds to given targets.

a Z b Z

v0
2 v0
S
S 1

F
F
E T
E T

N N

Figure 5: Examples of trajectories that correspond to the minimal starting speed of the projectile for a given
starting point S and a given target T : (a) from the North Pole to the Equator; (b) from a starting point S
that is raised to the height of R/3 over the Pole, to the Equator.

Let the starting point S be located at the North Pole, and the target T on the Equator (Fig. 5(a)). In this
case both points S and T are located on the surface of the Earth, so their distances from the center of the
√ Earth’s radius R: r0 = rT = R. √
Earth are both equal to the Hence the distance lT of the target T from the
starting point S equals 2R, so that b = rT + lT = (1 + 2)R. For the minimal starting speed Eq. (6)
yields in this case v0 min = 0.9102 vcirc , where vcirc ≈ 7.9 km/s is the circular speed for the (hypothetical)
orbit whose radius equals the Earth’s radius (the 1st cosmic velocity).
At the target point T both ellipses, the desired trajectory (labeled 1 in Fig. 5(a)) and the envelope
bounding surface, have a common tangent. According to the optical property, a ray ET emitted from the
common focus E of these ellipses must be reflected at T by both curves toward their second foci (F and S
respectively). Therefore all three points (T , F , and S) lie on the same straight line T S.
The angle T SZ (or F SZ, which is the same) in this case equals 135◦ . The initial velocity v0 must be
directed along the bisectrix of this angle. Hence the angle between vector v0 and the upward vertical line
must equal 67.5◦ .
It is interesting to compare this trajectory 1 with the trajectory 2 (see Fig. 5(a)), which is traced by the
projectile launched with the same initial speed v0 at an angle of 45◦ . We note that on the surface of the

7
Earth the range of this projectile is smaller than in the preceding case, in which the projectile is fired with
the same starting speed at an angle of 67.5◦ , contrary to the situation on the “flat Earth.”
As another example, we consider a starting point S located at a height of one third the Earth’s radius
R over the North Pole, with the target still on the Equator (Fig. 5(b)). In this case r0 = 43 R, rT = R,
and lT = 35 R, so that b = 83 R. For the minimal starting speed in this case Eq. (6) yields v02 min = 23 vcirc
2
,
or v0 min = 0.8165 vcirc . From the triangle T SE we can see that the sine of angle T SE equals 3/5. The
initial velocity v0 must be directed along the bisectrix of the angle T SZ. Hence the angle between vector
v0 and the upward vertical line must equal 71.565◦ . The trajectory of this projectile is a portion ST of an
ellipse whose foci are located at E (the Earth’s center) and F . The latter point lies on the straight line ST
joining the starting point and the target.
The bounding ellipse and the trajectories, shown in Figure 5 and subsequent figures, are generated with
the help of the relevant simulation program of the software package [6].

6 Targets in the Northern and Southern Hemispheres

a Z b Z
v0
v0
S
S
F T
F
T

E 30
E 45

N N

Figure 6: Examples of trajectories that correspond to the minimal starting speed of the projectile for a given
starting point S at the North Pole and a given target T : (a) a target located at northern latitude 45◦ ; (b) at
northern latitude 30◦
.

Figures 6(a) and 6(b) show trajectories of missiles with minimal shooting speed aimed at targets located
at northern latitudes 45◦ and 30◦ , respectively. Again both points S and T are located on the surface of the
Earth, so that their distances from the center of the Earth are equal to the Earth’s radius R: r0 = rT = R.
For the target T at 45◦ latitude the distance lT of the target from the starting point S equals 2R sin 22.5◦ ,
so that b = rT + lT = 1.765 R. According to Eq. (6) the minimal starting speed v0 min = 0.744 vcirc . The
angle T SZ in this case equals 112.5◦ . The initial velocity v0 must be directed along the bisectrix of this
angle. Hence the angle between vector v0 and the upward vertical line must equal 56.25◦ .
Similarly, for the example in Fig. 6(b), r0 = rT = R. For the target T at 30◦ latitude the distance lT
of the target from the starting point S√ also equals R, so that b = rT + lT = 2R. According to Eq. (6),
the minimal starting speed is v0 min = 2/3 vcirc = 0.8165 vcirc . The angle T SZ in this case equals 120◦ .
Hence the angle between vector v0 and the upward vertical line must equal 60.0◦ .

8
 a b
Z Z

v0
v0
S
S

F
F
E 30
E 45

N T
N T

Figure 7: Examples of trajectories that correspond to the minimal starting speed of the projectile for a given
starting point S at the North Pole and a given target T : (a) a target located at southern latitude 30◦ ; (b) at
southern latitude 45◦ .

Figures 7(a) and 7(b) show trajectories (and the envelope bounding surfaces) of missiles with minimal
shooting speed aimed at targets located at southern latitudes of 30◦ and 45◦ , respectively. We leave the
calculation of the required shooting speeds and the firing angles for these cases to the reader. Correctness
of the calculations can be verified with the help of an orbit simulation program.[6]
We note that for a starting point located at the North Pole and a target on the surface of the Earth, the
minimal initial speed of the missile tends to the circular velocity vcirc ≈ 7.9 km/s (the speed of a satellite
orbiting the Earth in an extremely low circular orbit) as the target is moved towards the South Pole. In order
for a missile launched from one pole to hit a target located at the opposite pole, the shooting speed must be
at least a little bit greater than vcirc .

7 Envelope of the Parabolic Trajectories

Although there exist in the literature[1]–[5] several elegant analytical derivations of the “safety paraboloid”
for projectiles moving in a uniform gravitational field, here we present for completeness a simple geometri-
cal proof of this well-known result (see Fig. 1). The derivation is similar to that in Sec. III for the elliptical
shape of the envelope of the family of Keplerian ellipses.
All the parabolic trajectories of the family with given initial speed v0 have a common starting point S
and a common directrix AA′ (see Fig. 8). The locus of foci of all the parabolas is a circle whose center is
located at the starting point S, and whose radius is equal to the distance from S to point Z, which is the
maximum height reached by the projectile launched vertically. This locus is shown by the dashed circle in
Fig. 8. Compare it with the circular locus of foci of elliptic orbits in Figs. 2–7.
In all cases, independently of the initial speed imparted to the projectiles, the center of this circle coin-
cides with the starting point S, and its radius equals the distance from S to point Z. This is also true for the
case under consideration, namely, for the limiting case of a very small initial speed v0 ≪ vcirc , in which
the upper portions of ellipses near their apexes can be approximately regarded as parabolas (approximation
of a “flat Earth”).
To prove that the envelope of parabolic trajectories is a parabola, we choose an arbitrary target point T
below the line AA′ , and draw a circle with a center at T (see Fig. 9). Let the radius of this circle be equal

9
 A Z A‘
v0

Figure 8: The envelope of the parabolic trajectories traced in a uniform field by projectiles launched from
one point with equal speeds, and the locus of foci of these parabolas (dashed circle).

B
A Z v0 A‘
T
S F

Figure 9: Geometric construction for the proof of the parabolic shape of the “safety domain” boundary (see
text for details).

to the distance from T to line AA′ , so that the circle grazes this horizontal line. The focus of the parabolic
trajectory that passes through T must belong to this circle, because line AA′ is the directrix of this parabola.
At the same time this focus must lie on the dashed circle (the locus of all the foci). Again we examine three
possibilities:
1. The circle centered at T crosses the dashed circle. In this case there exist two foci, and hence two
parabolic trajectories of the family pass through the given target T .
2. The circle centered at T has no common points with the dashed circle. This means that a trajectory
passing through T does not exist: the given target T is out of reach from S for a projectile with the
given initial velocity v0 .
3. Finally, the circle centered at T grazes the dashed circle. In this case there exists a unique trajectory
of the family that passes through the given target T . This means that in this case the point T belongs
to the desired boundary (see Fig. 9). At this point T the trajectory grazes the boundary.
To see that this boundary is a parabola, we draw one more horizontal line whose distance from the
starting point S equals twice the upper height |SZ| of the projectile launched vertically (see Fig. 9). In case
the two circles graze one another, the distance |T B| from the target T to this line just equals the distance
ST . This proves that point T belongs to a parabola, whose directrix is the horizontal line passing through
B, and whose focus is located at the starting point S. Thus we have proved that the envelope of parabolic
trajectories is a parabola whose apex is point Z and whose focus is point S. We also note that this boundary
of the “safety domain” can be obtained from the parabolic trajectory of the projectile that launched from Z
horizontally, by raising this parabola as a whole through distance |SZ| (see Fig. 9).
The focus F of the parabolic trajectory that passes through T lies on the straight line joining the starting
point S and the target T (see Fig. 9). This can be easily seen from the optical property of a parabolic

10
mirror, according to which any light ray emitted from the focus is reflected in the direction of the axis of
the parabola. Since at point T both parabolas (the trajectory and the envelope) graze each other and have
a common tangent, the light rays from their foci (F and S respectively) are both reflected in the same
direction. This means that points S, F , and T lie on the same straight line.
The above-mentioned optical property of a parabolic mirror allows us to easily find the required direc-
tion of the firing speed v0 at the initial point S in order to hit a given target T . Indeed, a light ray directed
vertically upward is reflected at point T by the parabolic mirror in the direction of its focus F . Hence the
tangent to this parabola at point T (as well as the direction of vector v0 , Fig. 9) is the bisectrix of the angle
ZSF (or ZST ).

8 Summary
In this paper alternative derivations are presented for the boundary of the “safety domain” of elliptic orbits of
satellites (and ballistic trajectories of missiles) in a central Newtonian gravitational field, and for parabolic
trajectories of projectiles in a uniform gravitational field, when the projectiles are fired from one point with
the same initial speed in various directions. The derivations for the shape of the envelope surface are almost
entirely based on geometrical properties of ellipses and parabolas. The boundary surface can be useful
for solving practical problems of finding the minimal firing speed and the required initial direction of a
projectile that is to hit a given target from a given starting point. These applications are illustrated through
several examples, and supported by computer simulations.

References
[1] D. Donnelly, “The parabolic envelope of constant initial speed trajectories,” Am. J. Phys. 60, 1149–
1151 (1992).
[2] A.P. French, “The envelopes of some families of fixed-energy trajectories,” Am. J. Phys. 61, 805–810
(1993).
[3] A. G. Alenitsyn, E. I. Butikov, A. S. Kondratyev, Concise Handbook of Mathematics and Physics (CRC
Press, New York, 1997), pp. 241–242.
[4] M. Baće et al. “The envelope of projectile trajectories,” Eur. J. Phys. 23, 637–642 (2002).

[5] E. I. Butikov, “Comment on ‘The envelope of projectile trajectories,”’ Eur. J. Phys. 24, L5–L9 (2003).
[6] E. I. Butikov, Planets and Satellites (Physics Academic Software, American Institute of Physics,
1999). An updated version for the Windows operating system is available as a free download at
<http://butikov.faculty.ifmo.ru>.
[7] L. D. Landau and E. M. Lifschitz Mechanics. New York: Pergamon (1976).

[8] Ch. Kittel, W. D. Knight, M. A. Ruderman Mechanics, Berkeley Physics Course, v. 1, New York:
McGraw-Hill (1965).

[9] E. I. Butikov, Motions of Celestial Bodies, IOP Publishing Ltd. (2014).

11