Des. Codes Cryptogr.

(2007) 44:97–113
DOI 10.1007/s10623-007-9068-3

Blocking Sets in P G(r, q n )

F. Mazzocca · O. Polverino · L. Storme

Received: 3 November 2006 / Accepted: 20 March 2007 / Published online: 26 July 2007
© Springer Science+Business Media, LLC 2007

Abstract Let S be a Desarguesian (n − 1)-spread of a hyperplane
of P G(r n, q). Let 

and B̄ be, respectively, an (n − 2)-dimensional subspace of an element of S and a minimal
blocking set of an ((r − 1)n + 1)-dimensional subspace of P G(r n, q) skew to . Denote by
K the cone with vertex  and base B̄, and consider the point set B defined by
B = (K \
) ∪ {X ∈ S : X ∩ K  = ∅}
in the Barlotti–Cofman representation of P G(r, q n ) in P G(r n, q) associated to the (n − 1)-
spread S . Generalizing the constructions of Mazzocca and Polverino (J Algebraic Combin,
24(1):61–81, 2006), under suitable assumptions on B̄, we prove that B is a minimal block-
ing set in P G(r, q n ). In this way, we achieve new classes of minimal blocking sets and we
find new sizes of minimal blocking sets in finite projective spaces of non-prime order. In
particular, for q a power of 3, we exhibit examples of r -dimensional minimal blocking sets
of size q n+2 + 1 in P G(r, q n ), 3 ≤ r ≤ 6 and n ≥ 3, and of size q 4 + 1 in P G(r, q 2 ),
4 ≤ r ≤ 6; actually, in the second case, these blocking sets turn out to be the union of q 3
Baer sublines through a point. Moreover, for q an even power of 3, we construct examples
of minimal blocking sets of P G(4, q) of size at least q 2 + 2. From these constructions, we
also get maximal partial ovoids of the hermitian variety H (4, q 2 ) of size q 4 + 1, for any q a
power of 3.

Keywords Blocking set · Ovoid · Spread

F. Mazzocca (B) · O. Polverino

Dipartimento di Matematica, Seconda Università degli Studi di Napoli, Via Vivaldi 43,
I-81100 Caserta, Italy
e-mail: [email protected]

O. Polverino
e-mail: [email protected]
L. Storme
Department of Pure Maths and Computer Algebra, Ghent University, Krijgslaan 281-S22,
9000 Gent, Belgium
e-mail: [email protected]

123
98 Des. Codes Cryptogr. (2007) 44:97–113

AMS Classifications 05B25 · 51E20 · 51E21

1 Introduction

Let P G(r, q) denote the r -dimensional projective space associated with the (r + 1)
-dimensional vector space G F(q)r +1 over the finite field G F(q) with q elements. Here,
always q = p h , p prime, h ≥ 1. Following [10], a blocking set in P G(r, q), r ≥ 2, is
defined as a point set B intersecting every hyperplane and containing no line. A subspace
S of P G(r, q) meeting a blocking set B just in one point P is said to be tangent to B in
P. Moreover, a point P of B is said to be essential if B \ {P} is not a blocking set, that
is, if a tangent hyperplane to B in P exists. When all points of B are essential, no proper
subset of B is a blocking set and B is called minimal. It is straightforward to see that, if B is
a minimal blocking set in P G(r, q), then B is a blocking set in any projective subspace of
P G(r, q) containing it and the minimality is preserved. Finally, we say that a blocking set B
is h-dimensional if the subspace generated by B has dimension h; in particular, for h = 2,
we say that B is planar.
The terminology on blocking sets is unfortunately not yet standard, so slight variations
of the previous definitions can be found. For example, in [6], a blocking set in P G(r, q) is
defined as a 1-blocking set. Now, we will survey some results useful in what follows.
Baer subplanes and unitals in P G(2, q 2 ), as well as ovoids in P G(3, q), are examples of
extremal minimal blocking sets, in the sense of the following two classical results.

Result 1 (Bruen [9] for r = 2; Beutelspacher [6] for r > 2) The minimal possible size of a
√
blocking set B in a finite projective space P G(r, q), r ≥ 2, is q + q + 1 and the bound is
attained if and only if q is a square and B is a Baer subplane.

In the case r > 2, improved results have been obtained by Storme and Weiner [18].

Result 2 (Bruen and Thas [10]) Let B be a minimal blocking set in P G(r, q). Then we have
the following results:
√
• if r = 2, |B| ≤ q q + 1 and equality holds if and only if q is a square and B is a unital;
• if r = 3, |B| ≤ q
2 + 1 and equality holds if and only if B is an ovoid;
• if r ≥ 4, |B| < q r +1 + 1.

An ovoid of P G(3, q), q > 2, is a (q 2 + 1)-cap. An ovoid of P G(3, 2) is a complete

5-cap of P G(3, 2). Every plane of P G(3, q) intersects an ovoid in either one or q + 1 points.
Furthermore, every point of an ovoid belongs to a unique plane intersecting this ovoid in one
point. This shows that an ovoid is a minimal blocking set of P G(3, q).
The notion of ovoid can be generalized to a non-singular quadric Q of P G(r, q) and to
a non-singular hermitian variety H (r, q 2 ) of P G(r, q 2 ): it is a point set of Q or H (r, q 2 )
meeting exactly once every generator (i.e. maximal subspace) of Q or H (r, q 2 ), respectively.
An ovoid of a non-singular parabolic quadric Q(2r, q) in P G(2r, q) contains exactly
q r + 1 points and it is known that it can only exist if r = 2, 3 (Gunawardena and Moorhouse
[11] for q odd, Thas [21] for q even). The three-dimensional elliptic quadrics of Q(4, q)
are the classical ovoids of Q(4, q). Furthermore, Q(6, q) has ovoids if q = 3h , h ≥ 1, but
Q(6, q) has no ovoids if q > 3 prime [4]. Every ovoid of Q(2r, q), r = 2, 3, is a minimal
blocking set in P G(2r, q), and meets every hyperplane in 1 mod p points; p the characteristic

123
Des. Codes Cryptogr. (2007) 44:97–113 99

of G F(q) (Ball [3], Ball et al. [4]). So far, for n = 2, 3, ovoids of Q(2n, q) are the largest
known examples of minimal blocking sets in P G(2n, q).
An ovoid of H (r, q 2 ) has size q r +1 + 1 if r is even and q r + 1 if r is odd. In [21], it is
proved that H (2r, q 2 ), with r ≥ 2, has no ovoid. An ovoid of H (3, q 2 ) is said to be locally
hermitian with respect to one of its points R if it is the union of q 2 Baer sublines through R.
A partial ovoid of H (r, q 2 ) is a set of points of H (r, q 2 ) meeting every generator of H (r, q 2 )
in at most one point and it is said to be maximal if it is not a subset of a larger partial ovoid.
In this paper, by using the Barlotti–Cofman representation of a projective space of non-
prime order [5], we generalize the cone constructions described in [15,19] and we achieve new
classes and new sizes of minimal blocking sets in P G(r, q n ). Choosing ovoids of P G(3, q),
Q(4, q) and Q(6, q) as base of our cones, we also obtain some large blocking sets (see
Theorems 2, 4, 7–10). In particular, for q an even power of 3, we obtain in P G(4, q) minimal
blocking sets of size q 2 +1 different from ovoids of parabolic quadrics, and minimal blocking
sets of size at least q 2 + 2. Finally, for q a power of 3, we construct maximal partial ovoids
of H (r, q 2 ), 4 ≤ r ≤ 6, of size q 4 + 1, which are the union of Baer sublines through a point.

2 The constructions

An (n − 1)-spread of a projective space P G(m − 1, q) is a family S of mutually disjoint

subspaces of dimension n − 1 such that each point of P G(m − 1, q) belongs to an element
of S . It has been proved by Segre [17] that (n − 1)-spreads of P G(m − 1, q) exist if and
only if m = r n. If r > 2, an (n − 1)-spread S of P G(r n − 1, q) is said to be normal if it
induces an (n − 1)-spread in any subspace generated by two of its elements, i.e. if A, B ∈ S ,
then any element of S is either disjoint from T = A, B
or contained in T . We say that an
(n − 1)-spread S of P G(r n − 1, q) is Desarguesian if either r > 2 and S is normal, or r = 2
and the associated translation plane obtained by André–Bruck and Bose representation (see
[1,7,8]) is Desarguesian.
Let S be a Desarguesian (n − 1)-spread of
= P G(r n − 1, q) and define an S -subspace
of
as a subspace X of
whose points are partitioned by elements of S . If X is an S -sub-
space different from a spread element, then S induces on it a Desarguesian (n − 1)-spread
and its dimension is of type tn − 1, with 2 ≤ t ≤ r.
The (n − 1)-spread S determines a representation P G(S ) of P G(r − 1, q n ) in
P G(r n − 1, q) where points and t-subspaces of P G(r − 1, q n ), 0 < t < r − 1, corre-
spond to elements of S and S -subspaces of dimension (t + 1)n − 1 of
, respectively.

Remark 1 Every hyperplane H of
contains exactly one S -subspace of dimension (r − 1)

n − 1. This S -subspace is the union of all spread elements contained in H.

Now, embed
as a hyperplane in
 = P G(r n, q) and define a point-line geometry

r = r (
 ,
, S ) in the following way:
• the points of r are the points of
 \
and the elements of S ;
• the lines of r are the n-subspaces of
 intersecting
in an element of S and the lines
of P G(S );
• the point-line incidences are inherited from
and
 .
The incidence structure r is isomorphic to the projective space P G(r, q n ) and we say
that r is the Barlotti–Cofman representation of P G(r, q n ) (see [5]). The points of r in

 \
will be called affine. Note that a t-subspace of r is either a tn-subspace of
 meeting

123
100 Des. Codes Cryptogr. (2007) 44:97–113

in an S -subspace of dimension tn − 1, or a t-subspace of P G(S ), i.e. an S -subspace of

of dimension (t + 1)n − 1. In particular, P G(S ) is a hyperplane of r .

Let Y be a fixed element of S and let  = n−2 be a hyperplane of Y . Let   = (r −1)n+1

be an ((r − 1)n + 1)-subspace of
disjoint from . Also, denote by  = (r −1)n the
(r − 1)n-subspace intersection of   and
, and by T the intersection point of  and Y .
Let B̄ be a blocking set of   such that B̄ ∩  = {Q}, Q a point, with the following
property:
(α)  \ {T }  ⊂ B̄, for every line  of   through T.
Denote by K the cone with vertex  and base B̄, i.e.

K = P̄, 

P̄∈ B̄

and note that, since   ∩  = ∅, P̄, 

∩ P̄  , 
= , for any distinct points P̄, P̄  ∈ B̄.
Let B be the subset of r defined by

B = (K \
) ∪ {X ∈ S : X ∩ K  = ∅}

and note that if Q ∈ Y (i.e. T = Q), then |B| = q n−1 (| B̄| − 1) + 1 and B ∩ P G(S ) = {Y };
while if Q  ∈ Y , then |B| = q n−1 | B̄| + 1 and |B ∩ P G(S )| = q n−1 + 1.

Proposition 1 B is a blocking set of the projective space r .

Proof Let S(r −1)n be an (r − 1)n-dimensional subspace of
 not contained in
and skew to
. Then dim S(r −1)n , 
= r n−1 and, as   ∩ = ∅, it turns out that dim( S(r −1)n , 
∩   )
= (r − 1)n, i.e. S(r −1)n , 
∩   is a hyperplane of   . This implies that there exists a point
P̄ ∈ B̄ ∩ S(r −1)n , 
∩   and, as a consequence, S(r −1)n ∩ P̄, 
 = ∅; i.e. the cone K
intersects any (r − 1)n-subspace of
 defining an (r − 1)-subspace in r . On the other
hand, since Y ∈ B, the hyperplane P G(S ) of r intersects B; so B meets every hyperplane
of r .
Suppose now that B contains a line of r .
If Q = T , then B ∩ P G(S ) = {Y } and a line of r contained in B must be defined by
an n-dimensional subspace Sn of
 with Sn ∩
= {Y }. From Sn ⊂ B, we get that the line
 = Sn ∩   through Q is contained in B̄, a contradiction.
If Q  = T , then |B ∩ P G(S )| = q n−1 + 1 and a line contained in B must be defined
by an n-dimensional subspace Sn of
 such that Sn ∩
= X ∈ S , with X ∩ K  = ∅. If
X = Y and  is the line Sn ∩   , then we have  \ {T } ⊂ B̄, contradicting Property (α). If
X  = Y , then Sn = Sn , 
∩   is an n-dimensional subspace of   containing Q such that
Sn \
⊂ B̄. This means that each line of Sn through the point Q and not in
is contained
in B̄, a contradiction. 

We point out that the notations introduced in this section will be used for the rest of the
paper, even without explicitly recalling them.

2.1 Construction A

Suppose that B̄ is a minimal blocking set of   such that T = Q, i.e. suppose that  is
a tangent hyperplane of B̄ at the point Q. In this case, B = (K \
) ∪ {Y } and |B| =
|K \
| + 1 = q n−1 (| B̄| − 1) + 1.

123
Des. Codes Cryptogr. (2007) 44:97–113 101

Let S(r −1)n be a hyperplane of r different from P G(S ). Note that since   ∩  = ∅, if
S(r −1)n contains Y , then dim(S(r −1)n ∩   ) = (r − 2)n + 1 and, if S(r −1)n does not contain
Y , then dim( S(r −1)n , 
∩   ) = (r − 1)n.

Lemma 1 Let S(r −1)n be a hyperplane of r different from P G(S ). If S(r −1)n contains Y
and S̄(r −2)n+1 is the subspace S(r −1)n ∩   , then
|B ∩ S(r −1)n | = q n−1 |( B̄ ∩ S̄(r −2)n+1 ) \ {Q}| + 1. (1)
If S(r −1)n does not contain Y and S̄(r −1)n = S(r −1)n , 
∩   , then
|B ∩ S(r −1)n | = | B̄ ∩ S̄(r −1)n |. (2)
Also, equality (2) holds for any hyperplane S̄(r −1)n of   not through Q and for each hyper-
plane S(r −1)n of r contained in , S̄(r −1)n
.

Proof Equality (1) is straightforward; so assume that S(r −1)n does not contain Y . Then

B ∩ S(r −1)n = (K \
) ∩ S(r −1)n = ( P̄, 
∩ S(r −1)n ).
P̄∈ B̄

Since S(r −1)n ∩  = ∅, we have that dim( P̄, 

∩ S(r −1)n ) ≤ 0 and equality occurs if and
only if P̄ ∈ B̄ ∩ S̄(r −1)n . So we can write

B ∩ S(r −1)n = ( P̄, 
∩ S(r −1)n )
P̄∈ B̄∩ S̄(r−1)n

and since P̄, 

∩ P̄  , 
=  for every pair of distinct points P̄, P̄  ∈ B̄, equality (2)
follows.
Now, assume that S̄(r −1)n is a hyperplane of   not containing Q and let H = S̄(r −1)n , 
∩

. Since dim H = r n − 2, by Remark 1, there exists in H a unique S -subspace S(r −1)n−1 of
dimension (r − 1)n − 1. Also, since Q  ∈ S̄(r −1)n , the spread element Y is not contained in
S(r −1)n−1 and the (r −1)-dimensional subspaces of r contained in , S̄(r −1)n
and skew to
 are exactly those passing through S(r −1)n−1 . If S(r −1)n is one of these (r − 1)-dimensional
subspaces of r , then S(r −1)n , 
∩   = S̄(r −1)n and equality (2) occurs. 

Proposition 2 B is a minimal blocking set of r if and only if B̄ is a minimal blocking set

of   .

Proof If B is a minimal blocking set of r , then, by Lemma 1, B̄ is a minimal blocking set

of   . Now, suppose that B̄ is a minimal blocking set of   . As B ∩ P G(S ) = {Y }, Y is an
essential point of B. Let P be an affine point of B, i.e. P ∈ K \
and let P̄ be the unique
point of B̄ such that P ∈ P̄, 
. Since B̄ is a minimal blocking set of   , there exists an
((r − 1)n)-dimensional subspace S̄(r −1)n of   such that S̄(r −1)n ∩ B̄ = { P̄}. Now, consider
the hyperplane S̄(r −1)n , 
of
 and its intersection with
, say H = S̄(r −1)n , 
∩
.
As H is a hyperplane of
containing , by Remark 1, H contains an S -subspace S(r −1)n−1
of dimension (r − 1)n − 1 not containing Y . By Lemma 1, the join of S(r −1)n−1 and P is an
(r − 1)n-dimensional subspace S(r −1)n of
 meeting B exactly in the point P, i.e. S(r −1)n
defines an (r − 1)-dimensional subspace of r tangent to B at P, hence P is an essential
point of B. 

From a t-dimensional blocking set, we obtain via Construction A blocking sets generating
projective spaces of dimension at most t, as we prove in the following proposition.

123
102 Des. Codes Cryptogr. (2007) 44:97–113

Proposition 3 If B̄ is a t-dimensional blocking set of   , then dim B

≤ min{r, t}.
Proof Let t = B̄
(t ≤ (r − 1)n + 1) and let t−1 = t ∩
. Also, let P1 , P2 , . . . , Pt−1 be
t − 1 points of t−1 , different from Q, such that t−1 = Q, P1 , P2 , . . . , Pt−1
, and let X i
be the element of S containing Pi , for i = 1, . . . , t − 1. Then S ∗ = Y, X 1 , X 2 , . . . , X t−1

is an S -subspace of
of dimension nl − 1, where l ≤ min{t, r }. Let Snl = t , S ∗

and note
that Snl ∩
= S ∗ , i.e. Snl defines an l-dimensional subspace l of r . Since K ⊂ Snl , the
blocking set B is contained in l , and hence dim B
≤ l ≤ min{r, t}. 

Theorem 1 From a minimal t-dimensional blocking set B̄ in a projective space of order

q, it is possible to obtain via Construction A minimal r -dimensional blocking sets B in
r ∼= P G(r, q n ) for any r such that max{2, t−1
n + 1} ≤ r ≤ t.
Proof Let B̄ be a minimal t-dimensional blocking set, and let r and n be two integers
(r, n ≥ 2) such that max{2, t−1 
n + 1} ≤ r ≤ t. Let r = r (
,
, S ) be the Barlotti–
Cofman representation of P G(r, q n ). Let Y be a spread element of S , and let  and Q be a
hyperplane and a point of Y , respectively, with Q  ∈ . Let X 1 , X 2 , . . . , X r −1 be r − 1 dis-
tinct elements of S such that
= Y, X 1 , X 2 , . . . , X r −1
. Let P j ∈ X j for j = 1, . . . , r − 1
and let   be an ((r − 1)n + 1)-dimensional subspace of
 disjoint from  containing
Q, P1 , P2 , . . . , Pr −1 . Since t−1  
n + 1 ≤ r ≤ t, there exists a t-dimensional subspace t of 

through Q such that P j ∈ t ∩
= t−1 , for any j = 1, . . . , r − 1. This means that the

 S -subspace containing t−1 is
. Embed B̄ in t in such a way that B̄ ∩
= {Q}, let
smallest
K = P̄∈ B̄ P̄, 
and let B be the blocking set obtained via Construction A from the cone
K . Since K
= Y, t
and since the smallest S -subspace containing K
∩
= Y, t−1

is
, we have B
= r . 

Now, we describe some three-dimensional minimal blocking sets that can be obtained via
Construction A.
Example 1 Some families of minimal three-dimensional blocking sets of P G(3, q) of size
2q + 1, 3q + 1 (q > 2), 4q + 1 (q > 2, q even), 3q − 1 (q > 2) and kq + 1 (q > 2, q even,
and 2 ≤ k ≤ q − 1) are constructed in [20] and are presented in [15, Sect. 3] as blocking
sets of type B̄1 , B̄2 , B̄3 , B̄4 and B̄5 , respectively. Choosing such blocking sets as base of the
cone K , by Theorem 1, we can construct minimal three-dimensional blocking sets B1 , B2 ,
B3 , B4 and B5 of P G(3, q n ), for any n ≥ 2, of size 2q n + 1, 3q n + 1 (q > 2), 4q n + 1
(q > 2, q even), 3q n − 2q n−1 + 1 (q > 2) and kq n + 1 (q > 2, q even, and 2 ≤ k ≤ q − 1),
respectively. It is not difficult to verify that none of such blocking sets is a blocking set of
type B̄1 , B̄2 , B̄3 , B̄4 , or B̄5 in P G(3, q n ).
Example 2 In [12], Heim constructed minimal blocking sets in P G(d, q) (q = p h ), not
contained in a hyperplane, of size (d − 1)q − (d − 3)q/ p + 1, if h > 2, and of size
(d − 1)q − (d − 3)(q + 1)/2 + d − 1, if q = p is an odd prime. Applying Construction A to
such blocking sets, we construct r -dimensional minimal blocking sets in P G(r, q n ) (q = p h ),
for any n ≥ 2 and r such that max{2, d−1
n +1} ≤ r ≤ d, of size (d −1)q −(d −3)q / p +1,
n n

if h > 2, and of size (d − 1)q − (d − 3)(q + q

n n n−1 )/2 + (d − 2)q n−1 + 1, if q = p is an
odd prime.

2.2 Ovoidal blocking sets of type A

We say that a blocking set B obtained via Construction A is an ovoidal blocking set of type
A if the base Ō of the corresponding cone is an ovoid either of P G(3, q) or of Q(4, q) or of
Q(6, q), and we write B = B Ō .

123
Des. Codes Cryptogr. (2007) 44:97–113 103

Ovoidal blocking sets of type A have a special structure as we show in the following
proposition.

Proposition 4 If B Ō is an ovoidal blocking set of type A in r ∼ = P G(r, q n ) with special

point Y , then each secant line through Y meets B Ō in q n−1 + 1 points. In particular, if n = 2,
then B Ō is a union of Baer sublines through Y .

Proof Let Sn be a line of r through Y not contained in P G(S ). Then we get

Sn ∩ B Ō = ∪ P̄∈Sn ∩ Ō ( , P̄
\ ) ∪ {Y }

and hence |Sn ∩ B Ō | = q n−1 |(Sn ∩ Ō) \ {Q}| + 1. Since Y ∩   = {Q}, Sn ∩   =  is a

line through Q. Hence, since Ō is an ovoid, we get |Sn ∩ B Ō | = 1 or q n−1 + 1. If n = 2 and
|S2 ∩ B Ō | = q + 1, then  ∩ Ō = { P̄, Q} and S2 ∩ B Ō = ( , P̄
\ ) ∪ {Y }, and such a
set is a Baer subline in the projective space r . 

Let Ō be an ovoid of P G(3, q). By Theorem 1, choosing Ō as base of the cone K , we

can construct three-dimensional minimal blocking sets B Ō in P G(3, q n ) of size q n+1 + 1
for any n ≥ 2. We call such ovoidal blocking sets of elliptic type if Ō is an elliptic quadric
and of Suzuki–Tits type if Ō is a Suzuki–Tits ovoid. We point out that an ovoidal blocking
set of elliptic type and an ovoidal blocking set of Suzuki–Tits type cannot be isomorphic. By
Lemma 1, it is easy to verify that a secant plane meets B Ō either in q + 1 or in q n−1 + 1 or
in q n + 1 points.
Let Ō be a non-classical ovoid of Q(4, q), q = p h . By Ball [3], Ō is a minimal blocking
set of P G(4, q) of size q 2 + 1 and hence, by Theorem 1, choosing Ō as base of the cone
K , we can obtain minimal blocking sets of size q n+1 + 1 in P G(3, q n ) and in P G(4, q n )
generating the whole space, for any n ≥ 2. Since there are four known families of non-
classical ovoids of Q(4, q) [13,22,23], we obtain with our construction four non-isomorphic
families of ovoidal blocking sets of size q n+1 + 1 of type A in P G(3, q n ) and in P G(4, q n )
generating the whole space, for any n ≥ 2. We also mention that there is a sporadic ovoid
in Q(4, 35 ) which can be used for Construction A [16]. It is not difficult to verify that, by
Lemma 1 and by Ball [3], any secant hyperplane meets B Ō in 1 mod p points.
Some of the above results can be summarized in the following theorem.

Theorem 2 In P G(3, q n ) and in P G(4, q n ), q = p h , p prime, h ≥ 1, for any n ≥ 2,

there exist non-isomorphic families of minimal blocking sets of size q n+1 + 1, generating
the whole space and intersecting each hyperplane in 1 mod p points. In particular, if n = 2,
such blocking sets are the union of q 2 Baer sublines through a point.

Consider the Barlotti–Cofman representation r = r (
 ,
, S ) of P G(r, q 2 ), r = 3, 4,

in
 = P G(2r, q) obtained from a normal line-spread S of
= P G(2r − 1, q) ⊂
 . A
hermitian variety H (r, q 2 ) of r having P G(S ) as tangent hyperplane at the point Y , Y ∈ S ,
is represented in
 by a quadratic cone K with vertex a point  of Y and base a hyperbolic
quadric Q + (5, q) if r = 3 and an elliptic quadric Q − (7, q) if r = 4. Also, K ∩
is a cone
with vertex the line Y which can be described as a union of elements of S .
Let Q(4, q) be a parabolic quadric contained in K intersecting Y in a point Q. If Ō is
an ovoid of Q(4, q) such that Ō ∩
= {Q} and K is the cone with vertex  and base Ō,
then K ⊂ K and the corresponding ovoidal blocking set B Ō of r is contained in H (r, q 2 ),
r = 3, 4. Note that, if Ō is a non-classical ovoid of Q(4, q), by Theorem 1, we can embed
Ō in
 in such a way that B Ō generates the whole space. Actually, if r = 3, B Ō is a locally

123
104 Des. Codes Cryptogr. (2007) 44:97–113

hermitian ovoid of H (3, q 2 ) with respect to the point Y corresponding to the symplectic
spread of P G(3, q) associated, via the inverse of the Plücker map, to the ovoid Ō, and any
locally hermitian ovoid of H (3, q 2 ) associated with a symplectic spread of P G(3, q) can be
obtained in this way (see [2,14] for more details). So we have proved the following theorem.

Theorem 3 Any locally hermitian ovoid O of H (3, q 2 ), q = p h , p prime, h ≥ 1, associated

with a symplectic spread of P G(3, q) is an ovoidal blocking set of type A of P G(3, q 2 ) and
every plane meets O in 1 mod p points.

In the four-dimensional case, we have the following result.

Theorem 4 In P G(4, q 2 ), q = p h , p an odd prime, h > 1, or q = 22e+1 , e ≥ 1, there

exist maximal partial ovoids of H (4, q 2 ) of size q 3 + 1 which are the union of Baer sublines
through a point, generating the whole space and intersecting each hyperplane in 1 mod p
points.

Proof In the above construction, let Ō be a non-classical ovoid of Q(4, q). Since | Ō| =
q 2 + 1, the blocking set B Ō has size q 3 + 1 and, by Proposition 4, it is the union of q 2 chords
(Baer sublines contained in H (4, q 2 )) through Y . Let R and S be two distinct points of B Ō
and suppose that the line π RS of 4 through R and S is a generator of H (4, q 2 ). Since B Ō is
a union of chords of H (4, q 2 ) through Y , the line π RS does not contain Y and hence π RS is a
plane of
 contained in the quadratic cone K intersecting
in a line of S . Let R  , S  be the
points of Ō such that R  ∈ , R
and S  ∈ , S
. Since the line of
 joining R and S is
contained in K, the plane π = , R, S
is contained in K and hence the line joining R  and
S  in
 is contained in Q(4, q), a contradiction. Hence, two points of B Ō are not contained
in a generator of H (4, q 2 ), i.e. B Ō is a partial ovoid of H (4, q 2 ).
Now suppose, by way of contradiction, that the partial ovoid B Ō is not maximal. This
means that there exists a point P ∈ H (4, q 2 ) \ B Ō such that B Ō ∪ {P} is a partial ovoid of
H (4, q 2 ). Then the tangent hyperplane to H (4, q 2 ) at the point P does not contain any point
of the blocking set B Ō , a contradiction. 

Finally, let Ō be an ovoid of Q(6, q). By Ball and Govaerts [4], Ō is a six-dimensional
minimal blocking set of P G(6, q) of size q 3 +1, hence, by choosing Ō as base of the cone K ,
we obtain a family of minimal blocking sets of size q n+2 + 1 of P G(r, q n ), with 3 ≤ r ≤ 6
if n ≥ 3, and of P G(r, q 2 ), with 4 ≤ r ≤ 6, generating the whole space. Since the known
families of ovoids of Q(6, q) are the Thas–Kantor ovoids with q = 3h and h ≥ 1, and
the Ree–Tits ovoids with q = 32h+1 , h > 0 (see for instance [13,21]), we obtain with our
construction two non-isomorphic families of ovoidal blocking sets of type A in characteristic
3. Also in this case, by Lemma 1 and by Ball and Govaerts [4], a secant hyperplane meets
B Ō in 1 mod p points. Hence, we have

Theorem 5 In P G(r, q n ), for 3 ≤ r ≤ 6, n ≥ 3 and q a power of 3, there exist minimal

blocking sets of size q n+2 + 1 generating the whole space and intersecting each hyperplane
in 1 mod three points. Also, there exist in P G(r, q 2 ), 4 ≤ r ≤ 6, q a power of 3, minimal
blocking sets of size q 4 + 1 generating the whole space, and which are the union of q 3 Baer
sublines through a point.

In some cases the ovoidal blocking sets of P G(r, q 2 ) obtained from ovoids of Q(6, q)
can be embedded in a hermitian variety of P G(r, q 2 ).

123
Des. Codes Cryptogr. (2007) 44:97–113 105

Consider the Barlotti–Cofman representation r = r (
 ,
, S ) of P G(r, q 2 ), 4 ≤

r ≤ 6, in
 = P G(2r, q) obtained from a normal line-spread S of
= P G(2r − 1, q) ⊂

 . A hermitian variety H (r, q 2 ) of r having P G(S ) as tangent hyperplane at the point Y ,
Y ∈ S , is represented in
 by a quadratic cone K with vertex a point  of Y and base an
elliptic quadric Q − (7, q) of
 if r = 4, a hyperbolic quadric Q + (9, q) if r = 5, and an
elliptic quadric Q − (11, q) if r = 6. Also, K ∩
is a cone with vertex the line Y which can
be described as a union of elements of S .
Let Q(6, q) be a parabolic quadric contained in K, and intersecting Y in a point Q. If Ō
is an ovoid of Q(6, q) such that Ō ∩
= {Q} and K is the cone with vertex  and base Ō,
then K ⊂ K and the ovoidal blocking set B Ō of r obtained from the cone K is contained
in H (r, q 2 ). Also in this case, by Theorem 1, we can embed Ō in
 in such a way that B Ō
generates the whole space and we have the following result.

Theorem 6 The minimal blocking set B Ō is a maximal partial ovoid of H (r, q 2 ), 4 ≤ r ≤ 6,

of size q 4 + 1 which is the union of q 3 Baer sublines through a point Y , generating the whole
space and intersecting each hyperplane in 1 mod p points.

Proof The result follows as in the proof of Theorem 4. 

Corollary 1 There exist maximal partial ovoids of H (r, q 2 ), 4 ≤ r ≤ 6 and q a power of 3,

of size q 4 + 1 which are the union of q 3 Baer sublines through a point.

2.3 Construction B

Suppose that Q  ∈ Y , let Z be the unique element of S such that Q ∈ Z and let  ∩ Y = {T }.
In this case, the size of B is given by
|B| = q n−1 (| B̄| − 1) + q n−1 + 1 = q n−1 | B̄| + 1
also B ∩ P G(S ) is contained in the line Y, Z
of P G(S ) and |B ∩ P G(S )| = q n−1 + 1.
The intersection numbers of B with respect to the hyperplanes can be determined as in
Construction A.
Recall that if S(r −1)n is a hyperplane of r other than P G(S ) containing Y , then
dim(S(r −1)n ∩   ) = (r − 2)n + 1. Whereas if S(r −1)n is a hyperplane of r not containing
Y , then dim( S(r −1)n , 
∩   ) = (r − 1)n.

Lemma 2 Let S(r −1)n be a hyperplane of r other than P G(S ). If S(r −1)n contains Y and
S̄(r −2)n+1 denotes the ((r − 2)n + 1)-dimensional subspace through T obtained intersecting
S(r −1)n and   , then
|B ∩ S(r −1)n | = q n−1 | S̄(r −2)n+1 ∩ B̄| + 1. (3)
If S(r −1)n does not contain Y and S̄(r −1)n = S(r −1)n , 
∩   , then
|B ∩ S(r −1)n | = | B̄ ∩ S̄(r −1)n |. (4)

Also, equality (4) holds for any hyperplane S̄(r −1)n of not through T and for each of the
q n−1 hyperplanes S(r −1)n of r contained in , S̄(r −1)n
.

Proof Let S(r −1)n be a hyperplane of r , other than P G(S ), containing Y and let S̄(r −2)n+1 =
S(r −1)n ∩   . Then we have
B ∩ S(r −1)n = ((K \
) ∩ S(r −1)n ) ∪ {X ∈ S : X ∩ K ∩ S(r −1)n  = ∅}.

123
106 Des. Codes Cryptogr. (2007) 44:97–113

Since

(K \
) ∩ S(r −1)n = ( P̄, 
\ )
P̄∈( B̄∩ S̄(r−2)n+1 )\{Q}

and |{X ∈ S : X ∩ K ∩ S(r −1)n  = ∅}| = 1 if Q  ∈ S̄(r −2)n+1 and q n−1 + 1 otherwise, we get
equality (3). If S(r −1)n is a hyperplane of r not containing Y , with the same arguments of
Lemma 1, we obtain the result. 

Remark 2 Note that if S(r −1)n is a hyperplane of r passing through a point of B ∩ P G(S )
different from Y , then Q ∈ S̄(r −1)n = S(r −1)n , 
∩   .

Proposition 5 Let P be an affine point of B and let { P̄} = P, 

∩   . The point P is an
essential point of B if and only if there exists in   a tangent hyperplane to B̄ at the point P̄,
not passing through T .
Proof If S(r −1)n is a hyperplane of r tangent to B at the point P, then by Lemma 2, S̄(r −1)n =
S(r −1)n , 
∩  is a hyperplane, not through T , of   tangent to B̄ at the point P̄. Conversely,
if there exists a tangent hyperplane S̄(r −1)n , not through T , to B̄ at the point P̄, by Lemma
2, there exist at least q n−1 hyperplanes of r tangent to B at the point P. 

The above proposition shows that the minimality of B does not automatically follow from
the minimality of B̄, as in Construction A; to this end, we need some extra conditions on B̄.
We say that B̄ satisfies Condition (∗) with respect to the point T if:
(∗) for each point P̄ ∈ B̄ \ {Q}, there exists a tangent hyperplane to B̄ passing through P̄,
but not containing T .
As a corollary of Proposition 5, we get

Corollary 2 The affine points of B are essential points if and only if B̄ satisfies Condition
(∗) with respect to the point T .
In the next proposition, we prove that if B̄ does not span the whole space   , then it is
possible to embed B̄ in   in such a way that all affine points of B are essential points.

Proposition 6 If B̄ is a minimal blocking set of   contained in a t-dimensional subspace

t of   , with t ≤ (r − 1)n and T  ∈ t , then all affine points of B are essential points.
Proof Since T  ∈ t , for each point P̄ ∈ B̄ \ {Q}, there exists a hyperplane of   tangent to
B̄ at the point P̄ which does not contain T , i.e. B̄ satisfies Condition (∗) with respect to the
point T . 

Now, let n be the n-dimensional subspace of  obtained intersecting Y, Z
and . Let X
be a point of B ∩ P G(S ) different from Y and let n−1
X = X, 
∩ n . It is clear that n−1
X is

a hyperplane of n through Q not containing the point T . By Lemma 2, the intersection num-
bers of B with respect to the hyperplanes of r through X , not containing Y , are determined
by the intersection numbers of B̄ with respect to the hyperplanes of   , different from ,
containing n−1
X . Indeed, if S
(r −1)n contains X and does not contain Y , then S̄(r −1)n contains
n−1 . Conversely, if n−1 is a hyperplane of n passing through Q and not containing T ,
X

then the intersection numbers of B̄ with respect to the hyperplanes of   , different from ,
containing n−1 , determine the intersection numbers of B with respect to the hyperplanes
of r (not containing Y ) through the unique element X of S \ {Y } contained in n−1 , 
.
Hence, we have:

123
Des. Codes Cryptogr. (2007) 44:97–113 107

Proposition 7 Let X ∈ B ∩ P G(S ), with X  = Y . The point X is an essential point of B

if and only if there exists a hyperplane of   , different from , tangent to B̄ at the point Q
and containing the subspace n−1 X = X, 
∩ n . Also, the number of essential points of B
on P G(S ) \ {Y } equals the number of hyperplanes of n passing through the point Q, not
containing the point T and contained in a tangent hyperplane to B̄ different from .

Denote by t the t-dimensional subspace spanned by B̄ and let t−1 = t ∩ . Since B̄

is contained in t , it is a blocking set of t with respect to the hyperplanes. From now on,
we suppose that T  ∈ t−1 (hence t ≤ (r − 1)n), so by Proposition 6, all the affine points of
B are essential points.
Let r −2 = S(r −1)n−1 be a hyperplane of P G(S ) through Y , not containing Z , let
 r−2 = S(r −1)n−1 ∩ t−1 , and note that dim  r−2 ≥ t − n − 1.

Proposition 8
(•) If there exists a hyperplane r −2 = S(r −1)n−1 of P G(S ) through Y , but not containing
Z , such that dim  r−2 ≥ t − n, then Y is an essential point of B. In particular, if t ≤ n
and t−1  ⊆ n , then Y is an essential point of B.
(••) If there exists a hyperplane r −2 = S(r −1)n−1 of P G(S ) through Y , not containing Z ,
such that dim  r−2 = t − n − 1 and such that  r−2 is contained in a (t − n)-dimensional
subspace t−n of t , not contained in
, and disjoint from B̄, then Y is an essential point
of B.

Proof
(•) Let r −2 = S(r −1)n−1 be the hyperplane of P G(S ) such that dim  r−2 = s ≥ t − n
and let r −1 = S(r −1)n be a hyperplane of r , different from P G(S ), containing r −2 .
By Lemma 2, since B̄ ⊂ t and since S(r −1)n ∩ t ∩
=  r−2 , we have |r −1 ∩ B| =
q n−1 |S r−1 ∩ B̄| + 1, where S r−1 = S(r −1)n ∩ t and s ≤ dim S r−1 ≤ s + 1. If for
all r −1 different from P G(S ) containing r −2 , the dimension of S r−1 is s + 1, then
|t \
| = q t = q n · q s+1 , and hence s = t − n − 1, a contradiction. Then there exists
a hyperplane  ¯ r −1 = S̄(r −1)n of r containing r −2 such that dim( S̄(r −1)n ∩ t ) = s,
i.e. S̄(r −1)n ∩ t =  r−2 and hence S r−1 ∩ B̄ = ∅. This means that  ¯ r −1 is a tangent
hyperplane to B at the point Y .
If t ≤ n and t−1  ⊆ n , it is possible to find a hyperplane r −2 of P G(S ) through Y ,
but not containing Z , such that dim  r−2 ≥ 0 ≥ t − n. So the preceding arguments show
that Y is an essential point of B.
(••) In this case, the hyperplane r −1 of r obtained joining r −2 and t−n  is a tangent
hyperplane to B at the point Y . 

If B  is a minimal blocking set of r contained in B, by Corollary 1, Propositions 7 and
8, we get

Corollary 3 If B̄ satisfies Condition (∗) with respect to the point T, then

|B  | = q n−1 (| B̄| − 1) + t Q + ε,
where t Q is the number of hyperplanes of n passing through Q, not containing T and con-
tained in a tangent hyperplane to B̄ different from ; whereas ε = 1 if Y is an essential point
of B and ε = 0 otherwise. Also, if t ≤ n, then ε = 1.

123
108 Des. Codes Cryptogr. (2007) 44:97–113

Proposition 9
(1) If t ≤ (r − 2)n + 1 and n ∩ t−1 = {Q}, then each point X ∈ B ∩ P G(S ), different
from Y , is an essential point of B. Hence, |B  | = q n−1 | B̄| + ε, where ε = 0 or 1.
(2) If (r − 2)n + 1 < t ≤ (r − 1)n and dim(n ∩ t−1 ) = t − (r − 2)n − 1, then the
number of essential points of B on P G(S ), different from Y , is at least q (r −1)n−t . Hence,
|B  | ≥ q n−1 (| B̄| − 1) + q (r −1)n−t + ε, where ε = 0 or 1.
Proof
(1) Let t ≤ (r − 2)n + 1. In this case, we may assume that n ∩ t−1 = {Q}. Let X ∈
B ∩ P G(S ) be different from Y and let n−1
X = X, 
∩n (recall that n = ∩ Y, Z
).
Then
X
dim n−1 , t−1
≤ (r − 1)n − 1
X ,
and T  ∈ n−1 t−1
. Hence, there exists a hyperplane (r −1)n−1 of  not containing
T and containing n−1 X ,  
t−1
. Now, if P ∈  \ (
∪ t ), then the space joining

(r −1)n−1 and P is a hyperplane of  , different from , tangent to B̄ at the point Q,
and containing the subspace n−1 X , thus, by Proposition 7, X is an essential point of B.

Hence, since T  ∈ t , all points of B \ {Y } are essential points, i.e. |B  | = q n−1 | B̄| + ε,


where ε = 0 or 1.
(2) Now, let t > (r −2)n +1, let dim(n ∩t−1 ) = t −(r −2)n −1, put s = t −(r −2)n −1
and s = n ∩ t−1 . If n−1 is a hyperplane of n not containing T and containing s ,
then n−1 , t−1
is a hyperplane of  not containing T and containing n−1 . Now, as
in the previous case, since t ≤ (r − 1)n, there exists a hyperplane of   , different from ,
tangent to B̄ at Q and containing n−1 . Hence, by Proposition 7, the number of essential
points of B on P G(S ) \ {Y } is at least the number of hyperplanes of n containing s ,
but not containing T . Hence, |B  | ≥ q n−1 (| B̄| − 1) + q (r −1)n−t + ε, where ε = 0 or 1.


Proposition 10 Suppose that there exists a hyperplane ¯ t−1 , different from t−1 , of t
tangent to B̄ at the point Q and let ¯ t−2 =
∩ ¯ t−1 . Then the following results hold.
(1 ) If t ≤ (r − 2)n + 2 and n ∩ ¯ t−2 = {Q}, then each point X ∈ B ∩ P G(S ), different
from Y , is an essential point of B. Hence, |B  | = q n−1 | B̄| + ε, where ε = 0 or 1.
(2 ) If (r − 2)n + 2 < t ≤ (r − 1)n and dim(n ∩ ¯ t−2 ) = t − (r − 2)n − 2, then the


number of essential points of B on P G(S ), different from Y , is at least q (r −1)n−t+1 .

Hence, |B  | ≥ q n−1 (| B̄| − 1) + q (r −1)n−t+1 + ε, where ε = 0 or 1.
Proof
(1 ) Let t ≤ (r − 2)n + 2 and suppose that n ∩ ¯ t−2 = {Q}. Let X ∈ B ∩ P G(S ) be
different from Y and let n−1 X = X, 
∩ n . As in case (1), there exists a hyperplane
(r −1)n−1 of  not containing T and containing n−1 X ,¯ t−2
. Now, the space joining
(r −1)n−1 and ¯ t−1 is a hyperplane of  , different from , tangent to B̄ at the point Q


and containing the subspace n−1 X , thus, by Proposition 7, X is an essential point of B.

Hence, |B  | = q n−1 | B̄| + ε, where ε = 0 or 1.

(2 ) Now, let t > (r −2)n +2, let dim(n ∩ ¯ t−2 ) = t −(r −2)n −2, put s  = t −(r −2)n −2
and s  = n ∩ ¯ t−2 . If n−1 is a hyperplane of n not containing T and containing
s  , then
dim n−1 , ¯ t−2
= n − 1 + t − 2 − s  = (r − 1)n − 1

123
Des. Codes Cryptogr. (2007) 44:97–113 109

and, since T  ∈ n−1 , T  ∈ n−1 , ¯ t−2

. Hence, there exists a hyperplane (r −1)n−1
of  not containing T and containing n−1 X , ¯ t−2
. Now, using the same arguments of

case (1) , we obtain the desired result. 

From a given t-dimensional blocking set, we can obtain via Construction B blocking sets
generating subspaces of dimension at most t + 1, as we prove in the following proposition.

Proposition 11 If B̄ is a t-dimensional blocking set, then dim B

≤ min{r, t + 1}.

Proof Let t = B̄
, t ≤ (r − 1)n + 1, and let t−1 = t ∩
. Also, let P1 , P2 , . . . , Pt be t
points of t−1 such that t−1 = P1 , P2 , . . . , Pt
and let X i be the element of S containing
Pi , for i = 1, . . . , t. Then S ∗ = Y, X 1 , X 2 , . . . , X t
is an S -subspace of
of dimension
nl − 1, where l ≤ min{t + 1, r }. Let Snl = t , S ∗
and note that Snl ∩
= S ∗ , i.e. Snl
defines an l-dimensional subspace, say l , of r . Also, since K ⊂ Snl , the blocking set B
is contained in l , i.e. dim B
≤ l ≤ min{r, t + 1}. 

In the next theorem, we determine the minimal blocking sets that can be obtained via
Construction B.

Theorem 7 From a minimal t-dimensional blocking set B̄ in a projective space of order q,

we can obtain via Construction B minimal r -dimensional blocking sets B  in P G(r, q n ) for
any r such that max{2, nt + 1} ≤ r ≤ t + 1. Also,
(a) |B  | = q n−1 | B̄| + 1 (i.e. B  = B) if t ≤ (r − 2)n + 1;
(b) |B  | = q n−1 | B̄| + ε (ε = 0, 1) if t = (r − 2)n + 2 and there exist at least two tangent
hyperplanes to B̄ at the point Q;
(c) |B  | ≥ q n−1 (| B̄| − 1) + q (r −1)n−t + ε (ε = 0, 1) if (r − 2)n + 1 < t ≤ (r − 1)n;
(d) |B  | ≥ q n−1 (| B̄| − 1) + q (r −1)n−t+1 + ε (ε = 0, 1) if (r − 2)n + 2 < t ≤ (r − 1)n and
there exist at least two tangent hyperplanes to B̄ at the point Q.

Proof Let B̄ be a minimal t-dimensional blocking set, and let r and n be two integers, r, n ≥ 2,
such that max{2, nt +1} ≤ r ≤ t +1. Let r be the r -dimensional projective space of order q n
obtained via the Barlotti–Cofman construction described in Sect. 2. Let Y, Z , X 1 , . . . , X r −2
be r elements of S such that
= Y, Z , X 1 , . . . , X r −2
, let T , Q, P j ( j = 1, . . . , r − 2) be
r points of
such that T ∈ Y , Q ∈ Z , and P j ∈ X j for j = 1, . . . , r − 2, and let  be a
hyperplane of Y not containing T . Also, let   be an ((r − 1)n + 1)-dimensional subspace
of
 disjoint from , containing T , Q, P1 , . . . , Pr −2 .
Since nt + 1 ≤ r ≤ t + 1, there exists a t-dimensional subspace t of   not contained
in
, such that Q, P j ∈ t ∩
= t−1 for j = 1, . . . , r − 2, and such that T  ∈ t . This

means that the smallest S -subspace  containing t−1 is
. Embed B̄ in t in such a way
that B̄ ∩
= {Q} and let K = P̄∈ B̄ P̄, 
. Since K
= , t
and since the smallest
S -subspace containing K
∩
= , t−1
is
, we have B
= r . By construction,
T  ∈ t , hence all the affine points of B are essential points, so if B  is the minimal blocking
set contained in B, we have that B
= B 
.
If t ≤ (r − 2)n + 1, we can choose t−1 in such a way that t−1 ∩ n = {Q} and hence, by
Proposition 9 case (1), |B  | = q n−1 | B̄| + ε (ε = 0, 1). Also, if t ≤ n, then ε = 1 (Proposition
8). If t > n and there exists a hyperplane r −2 of P G(S ) through Y , not containing Z , for
which dim  r−2 ≥ t − n, then Y is an essential point of B (Proposition 8 (•)). If t > n and
for all hyperplanes r −2 of P G(S ) through Y , not containing Z , dim  r−2 = t − n − 1, we
can embed B̄ in t in such a way that there exists a hyperplane r −2 = S(r −1)n−1 of P G(S )
through Y , not containing Z , such that dim  r−2 = t − n − 1 and  r−2 is contained in a

123
110 Des. Codes Cryptogr. (2007) 44:97–113

(t − n)-dimensional subspace t−n  of t , not contained in
, disjoint from B̄. So, by case

(••) of Proposition 8, |B | = q n−1 | B̄| + 1. This ends the discussion of case (a).
If (r − 2)n + 1 < t ≤ (r − 1)n, we can choose t−1 in such a way that dim(t−1 ∩ n ) =
t −[(r −2)n +1] and hence, by case (2) of Proposition 9, |B  | ≥ q n−1 (| B̄|−1)+q (r −1)n−t +ε
(ε = 0, 1). This treats case (c).
Now, suppose that there exists in t a tangent hyperplane ¯ t−1 to B̄ at the point Q different
from t−1 , and let ¯ t−2 = ¯ t−1 ∩ t−1 . In this case, if t ≤ (r − 2)n + 2, we can choose t−1
in such a way that ¯ t−2 ∩ n = {Q}, so by case (1) of Proposition 10, |B  | = q n−1 | B̄| + ε
(ε = 0, 1), leading to case (b). While, if (r − 2)n + 2 < t ≤ (r − 1)n, we can choose t−1 in
such a way that dim(¯ t−2 ∩ n ) = t − [(r − 2)n + 2] and hence by case (2) of Proposition
10, |B  | ≥ q n−1 (| B̄| − 1) + q (r −1)n−t+1 + ε (ε = 0, 1). This concludes the proof of case
(d). 

Applying Theorem 7, it is possible to construct a lot of examples of non-planar minimal

blocking sets, but we will only describe in detail the non-planar minimal blocking sets that
can be constructed starting from ovoids.

2.4 Ovoidal blocking sets of type B

We say that a blocking set B obtained via Construction B is an ovoidal blocking set of type
B if the base Ō of the corresponding cone is an ovoid either of P G(3, q) or of Q(4, q) or of
Q(6, q), and we write B = B Ō .

Proposition 12 Let B Ō be an ovoidal blocking set of type B in r ∼ = P G(r, q n ) with special

point Y , t ≤ (r − 1)n and T  ∈ t . Then each secant line through Y meets B Ō in q n−1 + 1
points. In particular, if n = 2, then B Ō is a union of Baer sublines through Y .

Proof Let Sn be a line of r through Y not contained in P G(S ). Then we get

Sn ∩ B Ō = ∪ P̄∈Sn ∩ Ō ( , P̄
\ ) ∪ {Y }

and hence |Sn ∩ B Ō | = q n−1 |Sn ∩ Ō| + 1. Since Y ∩ t = ∅, then |Sn ∩ t | ≤ 1 and so
|Sn ∩ B Ō | = 1 or q n−1 + 1. If n = 2 and |S2 ∩ B Ō | = q + 1, then S2 ∩ B Ō is a Baer subline
in the projective space r (see proof of Proposition 4). Finally, B Ō ∩ P G(S ) = B Ō ∩ Y, Z

and |B Ō ∩ Y, Z
| = q n−1 + 1. Also, if n = 2, B Ō ∩ Y, Z
is the set of lines of S intersecting
the line , Q
, i.e. B Ō ∩ Y, Z
is a Baer subline of r ∼ = P G(r, q 2 ). 

Let Ō be an ovoid of P G(3, q). By Theorem 7, choosing Ō as base of the cone K , we can
construct minimal blocking sets B Ō in P G(3, q n ) and P G(4, q n ), of size q n+1 + q n−1 + 1,
generating the whole space. By Lemma 2, it is easy to verify that a secant hyperplane meets
B Ō either in q + 1 or in q n−1 + 1 or in 2q n−1 + 1 or in q n + q n−1 + 1 points. Hence, the
following holds true.

Theorem 8 There exist in P G(3, q n ) and in P G(4, q n ) for any q and for any n ≥ 2, minimal
blocking sets of size q n+1 + q n−1 + 1 generating the whole space, with intersection numbers
with respect to the hyperplanes in the set {1, q + 1, q n−1 + 1, 2q n−1 + 1, q n + q n−1 + 1}. In
particular, if n = 2 such blocking sets are the union of q 2 + 1 Baer sublines through a point.

Now, let Ō be a non-classical ovoid of Q(4, q), q = p h . By Ball [3], Ō is a minimal

blocking set of P G(4, q) of size q 2 + 1 and hence, by Theorem 7, choosing Ō as base of the
cone K we can construct the following minimal blocking sets generating the whole space:

123
Des. Codes Cryptogr. (2007) 44:97–113 111

(1) B  = B Ō of size q n+1 + q n−1 + 1 in P G(3, q n ), P G(4, q n ) and P G(5, q n ) if n ≥ 3,

and in P G(4, q n ) and P G(5, q n ) if n = 2.
(2) B  of size q 3 + q + ε (ε = 0 or 1) in P G(3, q 2 ) if there exist at least two tangent
hyperplanes to Ō at the point Q, and B  of size at least q 3 + 1 + ε (ε = 0 or 1) in
P G(3, q 2 ) otherwise.
Applying the above constructions to the known examples of non-classical ovoids of
Q(4, q), we obtain the following result.

Theorem 9 In P G(r, q n ), 3 ≤ r ≤ 5 and n ≥ 3, and in P G(r, q 2 ), 4 ≤ r ≤ 5, with either

q = p h , p an odd prime and h > 1, or q = 22e+1 , e ≥ 1, there exist r -dimensional minimal
blocking sets of size q n+1 + q n−1 + 1 which are the union of (q n−1 + 1)-secants through
a point. In particular, if n = 2, such blocking sets are the union of q 2 + 1 Baer sublines
through a point.

Finally, let Ō be an ovoid of Q(6, q), q = p h . By Ball and Govaerts [4], Ō is a minimal
blocking set of P G(6, q) of size q 3 + 1 which spans the six-dimensional space, and hence,
by Theorem 7, choosing Ō as base of the cone K , we can construct the following minimal
blocking sets generating the whole space:
(a) B  = B Ō of size q n+2 + q n−1 + 1 in P G(r, q n ) for 4 ≤ r ≤ 7, if n ≥ 3, in P G(3, q n )
if n ≥ 5, and in P G(r, q n ) for 5 ≤ r ≤ 7, if n = 2 (case (a)).
(b) B  of size at least q n+2 + q 2n−6 + ε(ε = 0 or 1) in P G(3, q n ) if n = 3 or 4, and B  of
size at least q 4 + 1 + ε (ε = 0 or 1) in P G(4, q 2 ) (case (c)).
(c) B  of size q 4 + q + ε (ε = 0 or 1) in P G(4, q 2 ), and B  of size q 6 + q 3 + ε (ε = 0
or 1) in P G(3, q 4 ) if there exist at least two tangent hyperplanes to Ō at the point Q
(case (b)).
(d) B  of size at least q 5 + q + ε (ε = 0 or 1) in P G(3, q 3 ) if there exist at least two tangent
hyperplanes to Ō at the point Q (case (d)).
Since the known examples of ovoids of Q(6, q) are the Thas–Kantor ovoids, with q = 3h
and h ≥ 1, and the Ree–Tits ovoids, with q = 32h+1 , h > 0, the above constructions pro-
duce examples of minimal blocking sets in P G(r, q n ), 3 ≤ r ≤ 7, in projective spaces with
q = 3h and q = 32h+1 , h ≥ 1. In particular, if we rewrite case (a) for n = 2, we obtain the
following theorem.

Theorem 10 In P G(r, q), for 5 ≤ r ≤ 7, q = 32e , e ≥ 1, there exist minimal blocking sets
√
of size q 2 + q + 1 generating the whole space, and which are the union of Baer sublines
through a point.

Also, we can prove the following result.

Theorem 11 In P G(4, q), q = 32e , e ≥ 1, there exist minimal blocking sets of size at least
q 2 + 2.

Proof By (6) above, it suffices to prove that it is possible to apply Construction B to an ovoid
Ō of Q(6, q  ), q  = 3e , h ≥ 1, in such a way that Y is an essential point of B Ō . Since Ō is a
minimal blocking set of P G(6, q  ) = Q(6, q  )
, there exist four-dimensional subspaces of
P G(6, q  ) disjoint from Ō.
Let q = q 2 and consider the Barlotti–Cofman representation 4 = 4 (
 ,
, S ) of
P G(4, q) in
 = P G(8, q  ).

123
112 Des. Codes Cryptogr. (2007) 44:97–113

Let Y and Z be two different elements of the Desarguesian line spread S of
=

P G(7, q  ) ⊂
 , let  and T be two distinct points of Y and let Q be a point of Z . Let S5∗ be
a five-dimensional S -subspace of
containing Y and not containing Z , so S5∗ ∩ Y, Z
= Y .
Let 4 be a four-dimensional subspace of
 such that 4 ∩
= 3 is a three-dimensional
subspace of S5∗ disjoint from Y and let 1 be a line contained in Y, Z
passing through Q and
disjoint from Y . Note that S6 = 4 , S5∗
defines a hyperplane 3 of 4 . Also, 6 = 4 , 1

is a six-dimensional subspace of
 disjoint from Y containing Q. Let 7 = 6 , T

and
embed in 6 an ovoid Ō of a parabolic quadric Q(6, q  ) in such a way that Ō ∩
= {Q} and
4 ∩ Ō = ∅. Applying Construction B to the cone with vertex  and base Ō, we obtain an
ovoidal blocking set B Ō of 4 generating the whole space (see proof of Theorem 7). Since
4 ∩ Ō = ∅ and 4 ∩
= 3 ⊂ S5∗ , by case (••) of Proposition 8, Y is an essential point of
B Ō . If there exist at least two tangent hyperplanes to Ō at the point Q, then, by Theorem 7
√
case (b), B Ō is a minimal blocking set of P G(4, q) of size q 2 + q + 1. Otherwise, by The-
orem 7 case (c), B Ō contains a minimal blocking set of P G(4, q) of size at least q 2 + 2. 

Acknowledgments This work was supported by the Research Project of MIUR (Italian Office for University
and Research) “Strutture geometriche, combinatoria e loro applicazioni”, by the Mathematics Department of
the Seconda Università degli Studi di Napoli and by the Research group GNSAGA of INDAM. The third
author thanks the Fund for Scientific Research-Flanders, Belgium, for a research grant. Part of this research
was done when the third author was visiting the Seconda Università degli Studi di Napoli and the Università
degli Studi di Napoli “Federico II”. He wishes to thank the colleagues at these universities for their hospitality.

References

1. André J (1954) Über nicht-Desarguessche Ebenen mit transitiver translationsgruppe. Math Z 60:156–186
2. Bader L, Marino G, Polverino O, Trombetti R Spreads of P G(3, q) and ovoids of polar spaces. Forum
Math to appear
3. Ball S (2004) On ovoids of O(5, q). Adv Geom 4(1):1–7
4. Ball S, Govaerts P, Storme L (2006) On ovoids of parabolic quadrics. Des Codes Cryptogr 38(1):131–145
5. Barlotti A, Cofman J (1974) Finite sperner spaces constructed from projective and affine spaces. Abh
Math Semin Univ Hamb 40:231–241
6. Beutelspacher A (1980) Blocking sets and partial spreads in finite projective spaces. Geom Dedicata
9:425–449
7. Bruck RH, Bose RC (1964) The construction of translation planes from projective spaces. J Algebra
1:85–102
8. Bruck RH, Bose RC (1966) Linear representations of projective planes in projective spaces. J Algebra
4:117–172
9. Bruen AA (1970) Baer subplanes and blocking sets. Bull Am Math Soc 76:342–344
10. Bruen AA, Thas JA (1982) Hyperplane coverings and blocking sets. Math Z 181:407–409
11. Gunawardena A, Moorhouse GE (1997) The non-existence of ovoids in O9 (q). Eur J Combin 18:171–173
12. Heim U (1997) Proper blocking sets in projective spaces. Discrete Math 174:167–176
13. Kantor WM (1982) Ovoids and translation planes. Can J Math 34:1195–1207
14. Lunardon G (2006) Blocking sets and semifields. J Combin Theory A 113:1172–1188
15. Mazzocca F, Polverino O (2006) Blocking sets in P G(2, q n ) from cones of P G(2n, q). J Algebraic
Combin 24(1):61–81
16. Penttila T, Williams B (2000) Ovoids of parabolic spaces. Geom Dedicata 82(1–3):1–19
17. Segre B (1964) Teoria di Galois, fibrazioni proiettive e geometrie non desarguesiane. Ann Mat Pura Appl
64:1–76
18. Storme L, Weiner Zs (2000) Minimal blocking sets in P G(n, q), n ≥ 3. Des Codes Cryptogr 21:235–251
19. Szőnyi T, Cossidente A, Gács A, Mengyán C, Siciliano A, Weiner Zs (2005) On large minimal blocking
sets in P G(2, q). J Comb Designs 13:25–41
20. Tallini G (1991) Blocking sets with respect to planes in P G(3, q) and maximal spreads of a non-singular
quadric in P G(4, q). Mitt Math Semin Giessen 201:97–107
21. Thas JA (1981) Ovoids and spreads of finite classical polar spaces. Geom Dedicata 174:135–143

123
Des. Codes Cryptogr. (2007) 44:97–113 113

22. Thas JA, Payne SE (1994) Spreads and ovoids in finite generalized quadrangles. Geom Dedicata
52(3):227–253
23. Tits J (1962) Ovoides et groupes de Suzuki. Arch Math 13:187–198

123