1. Introduction

Throughout this paper, all groups are finite and \( G \) stands for a finite group. Moreover, \( 𝕇 \) is the set of all primes, \( \pi\subseteq 𝕇 \) and \( \pi^{\prime}=𝕇\setminus\pi \); and \( \pi(G) \) is the set of all primes dividing \( |G| \). Furthermore, \( Z_{\mathfrak{U}_{\pi}}(G) \) is the \( \pi \)-supersoluble hypercenter of \( G \), i.e., the product of all normal subgroups \( N \) of \( G \) such that every chief factor of \( G \) below \( N \) is either cyclic or a \( \pi^{\prime} \)-group, and \( Z_{\mathfrak{U}}(G)=Z_{\mathfrak{U}_{𝕇}}(G) \) is the supersoluble hypercenter of \( G \).

In what follows, \( \sigma \) is some partition of \( 𝕇 \), i.e., \( \sigma=\{\sigma_{i}\mid i\in I\} \), where \( 𝕇=\bigcup_{i\in I}\sigma_{i} \) and \( \sigma_{i}\cap\sigma_{j}=\varnothing \) for all \( i\neq j \); \( \sigma(G)=\{\sigma_{i}\mid\sigma_{i}\cap\pi(G)\neq\varnothing\} \) (see [1]).

A set \( {\mathcal{H}} \) of subgroups of \( G \) is said to be a complete Hall \( \sigma \)-set of \( G \) (see [1]) if each nonidentity member of \( {\mathcal{H}} \) is a Hall \( \sigma_{i} \)-subgroup of \( G \) for some \( i\in I \) and \( \mathcal{H} \) has exactly one Hall \( \sigma_{i} \)-subgroup of \( G \) for every \( i \).

A subgroup \( A \) of \( G \) is said to be \( \sigma \)-permutable in \( G \) (see [2]) if \( G \) possesses a complete Hall \( \sigma \)-set and \( A \) permutes with every Hall \( \sigma_{i} \)-subgroup \( H \) of \( G \), i.e., \( AH=HA \) for all \( i \) and \( A \) is \( \sigma \)-semipermutable in \( G \) [3] if \( G \) possesses a complete Hall \( \sigma \)-set \( {\mathcal{H}} \) such that \( AH^{x}=H^{x}A \) for all \( x\in G \) and all \( H\in{\mathcal{H}} \) with \( \sigma(A)\cap\sigma(H)=\varnothing \).

The theories of \( \sigma \)-permutable and \( \sigma \)-semipermutable subgroups are closely related to the theories of \( \sigma \)-soluble and \( \sigma \)-nilpotent groups [1,2,3,4,5].

Recall that \( G \) is said to be \( \sigma \)-decomposable (see [6]) or \( \sigma \)-nilpotent (see [2]) if \( G \) is \( \sigma_{i} \)-closed for all \( i \); \( \sigma \)-soluble (see [2]) if every chief factor \( H/K \) of \( G \) is a \( \sigma_{i} \)-group for some \( i \); and \( G^{\mathfrak{N}_{\sigma}} \) is the \( \sigma \)-nilpotent residual of \( G \), i.e., the smallest normal subgroup of \( G \) with \( \sigma \)-nilpotent quotient.

Let \( \tau_{\mathcal{H}}(A)=\{\sigma_{i}\in\sigma(G)\setminus\sigma(A)\mid\sigma(A)\cap\sigma(H^{G})\neq\varnothing \) for a Hall \( \sigma_{i} \)-subgroup \( H\in{\mathcal{H}} \)} (see [7]).

Then we say, following Beidleman and Skiba [7], that a subgroup \( A \) of \( G \) is as follows:

(i) \( \tau_{\sigma} \)-permutable in \( G \) with respect to \( {\mathcal{H}} \) if \( AH^{x}=H^{x}A \) for all \( x\in G \) and all \( H\in\mathcal{H} \) such that \( \sigma(H)\subseteq\tau_{\mathcal{H}}(A) \);

(ii) \( \tau_{\sigma} \)-permutable in \( G \) if \( A \) is \( \tau_{\sigma} \)-permutable in \( G \) with respect to some complete Hall \( \sigma \)-set \( \mathcal{H} \) of \( G \).

In the classical case when \( \sigma=\sigma{{}^{1}}=\{\{2\},\{3\},\dots\} \) (we use here the notations of [1]), the \( {\sigma} \)-permutable, \( \sigma \)-semipermutable, and \( \tau_{\sigma} \)-permutable subgroups are called respectively \( S \)-permutable (see [8]), \( S \)-semipermutable (see [9]), and \( \tau \)-permutable (see [10]).

Finally, recall that \( G \) is said to be a \( P\sigma T \)-group (see [2]) if \( \sigma \)-permutability is a transitive relation in \( G \); i.e., if \( H \) is a \( \sigma \)-permutable subgroup of \( K \) and \( K \) is a \( \sigma \)-permutable subgroup of \( G \), then \( H \) is \( \sigma \)-permutable in \( G \). In the case when \( \sigma=\sigma^{1} \), a \( P\sigma T \)-group is called a \( PST \)-group [8].

The theory of \( P\sigma T \)-groups was developed in [1, 2, 5, 11], and the following theorem is one of the culmination results of the theory.

Theorem A (see Theorem A in [1])

Let \( D=G^{\mathfrak{N_{\sigma}}} \). If \( G \) is a \( \sigma \)-soluble \( P\sigma T \)-group, then the following hold:

(i) \( G=D\rtimes M \), where \( D \) is an abelian Hall subgroup of \( G \) of odd order, \( M \) is \( \sigma \)-nilpotent, and every element of \( G \) induces a power automorphism in \( D \);

(ii) \( O_{\sigma_{i}}(D) \) has a normal complement in a Hall \( \sigma_{i} \)-subgroup of \( G \) for all \( i \).

Conversely, if (i) and (ii) hold for some subgroups \( D \) and \( M \) of \( G \), then \( G \) is a \( P\sigma T \)-group.


In this paper, basing on Theorem A and some results of [7], we obtain the following characterization of \( \sigma \)-soluble \( P\sigma T \)-groups:

Theorem B

\( G \) is a \( \sigma \)-soluble \( P\sigma T \)-group if and only if the following hold:

(i) \( G \) possesses a complete Hall \( \sigma \)-set \( {\mathcal{H}}=\{H_{1},\dots,H_{t}\} \) and a normal subgroup \( N \) with \( \sigma \)-nilpotent quotient \( G/N \) such that \( H_{i}\cap N\leq Z_{\mathfrak{U}_{\pi}}(H_{i}) \) for all \( i \), where \( \pi=\pi(N) \);

(ii) Every \( \sigma_{i} \)-subgroup of \( G \) is \( \tau_{\sigma} \)-permutable in \( G \) for all \( \sigma_{i}\in\sigma(N) \).


Since every \( \sigma \)-semipermutable subgroup is \( \tau_{\sigma} \)-permutable, we get from Theorem B the following already-known result:

Corollary 1.1 (see Theorem A in [3])

Let \( D=G^{{\mathfrak{N}}_{\sigma}} \) and \( \pi=\pi(D) \). Suppose that \( G \) possesses a complete Hall \( \sigma \)-set \( \mathcal{H} \) all members of which are \( \pi \)-supersoluble. If every \( \sigma_{i} \)-subgroup of \( G \) is \( \sigma \)-semipermutable in \( G \) for all \( \sigma_{i}\in\sigma(D) \), then \( G \) is a \( \sigma \)-soluble \( P\sigma T \)-group.


Note that Theorem B remains new for each special partition \( \sigma \) of \( 𝕇 \). In particular, in the case when \( \sigma=\sigma^{1} \) we get from Theorem B the following new characterization of the soluble \( PST \)-groups.

Corollary 1.2

Let \( D=G^{\mathfrak{N}} \) be the nilpotent residual of \( G \) and \( \pi=\pi(D) \). Then \( G \) is a soluble \( PST \)-group if and only if every \( p \)-subgroup of \( G \) is \( \tau \)-permutable in \( G \) for all \( p\in\pi \).


The proof of Theorem B consists of many steps and the following theorem is one of them.

Theorem C

Let \( D=G^{\mathfrak{N_{\sigma}}} \) and \( \pi=\pi(D) \). Suppose that \( G \) possesses a complete Hall \( \sigma \)-set \( {\mathcal{H}}=\{H_{1},\dots,H_{t}\} \) such that \( H_{i}\cap D\leq Z_{\mathfrak{U}_{\pi}}(H_{i}) \) for all \( i \). If all maximal subgroups of every noncyclic Sylow \( p \)-subgroup of \( G \) are \( \tau_{\sigma} \)-permutable in \( G \) for all \( p\in\pi \), then

(i) \( D \) is a nilpotent Hall subgroup of \( G \), \( D\leq Z_{\mathfrak{U}}(G) \);

(ii) \( (p-1,|G|)\neq 1 \) for every prime \( p \) dividing \( |D| \). Hence, \( p\in\pi(G/D) \) for the smallest prime \( p \) dividing \( |G| \).

Corollary 1.3 (see Theorem 10.3 in [12, VI])

If every Sylow subgroup of \( G \) is cyclic, then \( G \) is supersoluble.

Corollary 1.4 (see Theorem B in [3])

Let \( D=G^{{\mathfrak{N}}_{\sigma}} \) and \( \pi=\pi(D) \). Suppose that \( G \) possesses a complete Hall \( \sigma \)-set \( \mathcal{H} \) such that every member \( H \) of \( \mathcal{H} \) with \( H\cap D\neq 1 \) is \( \pi \)-supersoluble. If all maximal subgroups of every noncyclic Sylow \( p \)-subgroup of \( G \) are \( \sigma \)-semipermutable in \( G \) for all \( p\in\pi \), then \( D \) is a nilpotent Hall subgroup of \( G \) of odd order and every chief factor of \( G \) below \( D \) is cyclic.


The unexplained terminology and notation are standard. The reader is referred to [9, 12, 13] if need be.

2. Proof of Theorem C

We use \( \mathfrak{N}_{\sigma} \) to denote the class of all \( \sigma \)-nilpotent groups.

Lemma 2.1 [2, Corollary 2.4 and Lemma 2.5]

The class \( {\mathfrak{N}}_{\sigma} \) is closed under direct products, homomorphic images and subgroups. Moreover, if \( E \) is a normal subgroup of \( G \) and \( E/(E\cap\Phi(G)) \) is \( \sigma \)-nilpotent, then \( E \) is \( \sigma \)-nilpotent.


In view of Proposition 2.2.8 in [14], we get from Lemma 2.1 the following

Lemma 2.2

If \( N \) is a normal subgroup of \( G \), then \( (G/N)^{{\mathfrak{N}}_{\sigma}}=G^{{\mathfrak{N}}_{\sigma}}N/N \).

Lemma 2.3 [15]

Let \( H \), \( K \), and \( N \) be pairwise permutable subgroups of \( G \) and let \( H \) be a Hall subgroup of \( G \). Then \( N\cap HK=(N\cap H)(N\cap K) \).


Recall that \( G \) is a \( D_{\pi} \)-group if \( G \) possesses a Hall \( \pi \)-subgroup \( E \) and every \( \pi \)-subgroup of \( G \) lies in some conjugate of \( E \); a \( \sigma \)-full group of Sylow type (see [16]), if every subgroup \( E \) of \( G \) is a \( D_{\sigma_{i}} \)-group for every \( \sigma_{i}\in\sigma(E) \), and \( \sigma \)-full (see [16]), provided that \( G \) possesses a complete Hall \( \sigma \)-set.

In view of Theorems A and B in [16], the following is true:

Lemma 2.4

If \( G \) is \( \sigma \)-soluble, then \( G \) is a \( \sigma \)-full group of Sylow type.

Lemma 2.5 [2, Lemma 3.1]

Let \( H \) be a \( \sigma_{i} \)-subgroup of a \( \sigma \)-full group \( G \). Then \( H \) is \( \sigma \)-permutable in \( G \) if and only if \( O^{\sigma_{i}}(G)\leq N_{G}(H) \).

Lemma 2.6 [7, Lemma 2.6]

Suppose that \( G \) has a complete Hall \( \sigma \)-set \( {\mathcal{H}}=\{H_{1},\dots,H_{t}\} \) such that the subgroups \( H \) and \( K \) of \( G \) are \( \tau_{\sigma} \)-permutable in \( G \) with respect to \( {\mathcal{H}} \). Let \( R \) be a normal subgroup of \( G \) and \( H\leq L\leq G \). Then

(1) \( {\mathcal{H}}_{0}=\{H_{1}R/R,\dots,H_{t}R/R\} \) is a complete Hall \( \sigma \)-set of \( G/R \). Moreover, if \( \sigma(H)=\sigma(HR/R) \), then \( HR/R \) is \( \tau_{\sigma} \)-permutable in \( G/N \) with respect to \( {\mathcal{H}}_{0} \).

(2) If \( HK=KH \) and \( \sigma(H\cap K)=\sigma(H)=\sigma(K) \), then \( H\cap K \) is \( \tau_{\sigma} \)-permutable in \( G \) with respect to \( {\mathcal{H}} \).

(3) If \( H\leq O_{\sigma_{i}}(G) \) for some \( i \), then \( H \) is \( {\sigma} \)-permutable in \( G \).

(4) If \( G \) is a \( \sigma \)-full group of Sylow type, then \( H \) is \( \tau_{\sigma} \)-permutable in \( L \).

Lemma 2.7

Let \( Z=Z_{{\mathfrak{U}}_{\pi}}(G) \). Then

(1) each chief factor of \( G \) below \( Z \) is either cyclic or a \( \pi^{\prime} \)-group;

(2) \( Z\cap E\leq Z_{{\mathfrak{U}}_{\pi}}(E) \) for every subgroup \( E \) of \( G \);

(3) \( NZ/N\leq Z_{{\mathfrak{U}}_{\pi}}(G/N) \) for every normal subgroup \( N \) of \( G \).

Proof

(1): In fact, it suffices to prove that if \( A \) and \( B \) are normal subgroups of \( G \) such that each chief factor of \( G \) below \( A \) is either cyclic or a \( \pi^{\prime} \)-group and each chief factor of \( G \) below \( B \) is either cyclic or a \( \pi^{\prime} \)-group, then each chief factor \( H/K \) of \( G \) below \( AB \) is either cyclic or a \( \pi^{\prime} \)-group. Moreover, in view of the Jordan–Hölder Theorem for chief series, it suffices to show that if \( A\leq K<H\leq AB \), then \( H/K \) is either cyclic or a \( \pi^{\prime} \)-group. But this follows from \( H=A(H\cap B)=K(H\cap B) \) and the \( G \)-isomorphism \( K(H\cap B)/K\simeq(H\cap B)/(K\cap B) \). Therefore, each chief factor of \( G \) below \( Z \) is either cyclic or a \( \pi^{\prime} \)-group.

(2): Let \( 1=Z_{0}<Z_{1}<\cdots<Z_{t-1}<Z_{t}=Z \) be a chief series of \( G \) below \( Z \). Then each factor \( Z_{i}/Z_{i-1} \) of the series is either cyclic or a \( \pi^{\prime} \)-group by (1).

Consider the normal series

$$ 1=Z_{0}\cap E\leq Z_{1}\cap E\leq\cdots\leq Z_{t-1}\cap E\leq Z_{t}\cap E=Z\cap E $$

in \( E \). Assume that \( (Z_{i}\cap E)/(Z_{i-1}\cap E) \) is not a \( \pi^{\prime} \)-group. Then, in view of the isomorphism,

$$ (Z_{i}\cap E)/(Z_{i-1}\cap E)\simeq(Z_{i}\cap E)Z_{i-1}/Z_{i-1}\leq Z_{i}/Z_{i-1} $$

we get that \( Z_{i}/Z_{i-1} \) is cyclic, and so \( (Z_{i}\cap E)/(Z_{i-1}\cap E) \) is cyclic. Therefore, in view of the Jordan–Hölder Theorem, each chief factor of \( E \) below \( Z\cap E \) is either cyclic or a \( \pi^{\prime} \)-group. Hence \( Z\cap E\leq Z_{{\mathfrak{U}}_{\pi}}(E) \).

(3): Let \( (H/N)/(K/N) \) be a chief factor of \( G/N \) such that \( H/N\leq NZ/N \). Then, in view of the isomorphism \( (H\cap Z)K/K\simeq(H\cap Z)/(K\cap Z) \), we have that \( H/K=(H\cap Z)K/K \) is a chief factor of \( G \) such that \( H/K \) is either cyclic or a \( \pi^{\prime} \)-group by (1). Hence \( NZ/N\leq Z_{{\mathfrak{U}}_{\pi}}(G/N) \). The lemma is proved.

The following lemma is a corollary of Theorem 6.7 in [13, IV].

Lemma 2.8

Let \( N\leq E \) be normal subgroups of \( G \) such that \( N\leq\Phi(E) \) and every chief factor of \( G \) between \( E \) and \( N \) is cyclic. Then each chief factor of \( G \) below \( E \) is cyclic.


A group \( G \) is said to be \( \sigma \)-primary (see [2]) if \( G \) is a \( \sigma_{i} \)-group for some \( i \).

Lemma 2.9

Let \( D=G^{\mathfrak{N_{\sigma}}} \) and \( p\in\pi=\pi(D) \), where \( p \) is the smallest prime dividing \( |D| \). If all maximal subgroups of every Sylow \( p \)-subgroup of \( G \) are \( \tau_{\sigma} \)-permutable in \( G \), then \( D \) is \( p \)-soluble.

Proof

Suppose that this lemma is false and let \( G \) be a counterexample of minimal order. Then \( D\neq 1 \). Assume that \( p\in\sigma_{k} \).

We show first that \( DR/R\simeq D/(D\cap R) \) is \( p \)-soluble for every minimal normal subgroup \( R \) of \( G \). Indeed, in case \( p \) does not divide \( |DR/R| \), it is clear. Suppose that \( p\in\pi(DR/R) \). Then \( p \) is the smallest prime dividing \( |DR/R| \), where \( DR/R=(G/R)^{{\mathfrak{N}}_{\sigma}} \) by Lemma 2.2.

Let \( V/R \) be a maximal subgroup of a Sylow \( p \)-subgroup \( P/R \) of \( G/R \). Then \( P/R=G_{p}R/R \) and \( V=R(V\cap G_{p}) \) for some Sylow \( p \)-subgroup \( G_{p} \) of \( G \). Hence

$$ p=|(P/R):(V/R)|=|G_{p}R:R(V\cap G_{p})|=|G_{p}|:|V\cap G_{p}|=|G_{p}:(V\cap G_{p})|, $$

and so \( V\cap G_{p} \) is a maximal subgroup of \( G_{p} \). Therefore, \( V\cap G_{p} \) is \( \tau_{\sigma} \)-permutable in \( G \) by hypothesis, and so \( V/R=R(V\cap G_{p})/R \) is \( \tau_{\sigma} \)-permutable in \( G/R \) by Lemma 2.6(1). The choice of \( G \) implies that \( (G/R)^{{\mathfrak{N}}_{\sigma}}=DR/R\simeq D/(D\cap R) \) is \( p \)-soluble.

Hence \( R\leq D \) and \( R \) is nonabelian. It is easy to see that \( R \) is the unique minimal normal subgroup of \( G \) and \( C_{G}(R)=1 \). By [12, IV, Theorem 2.8], a Sylow \( p \)-subgroup \( Q \) of \( R \) is not cyclic. Hence \( |Q|>p \).

Let \( P \) be a Sylow \( p \)-subgroup of \( G \) such that \( Q=P\cap R \). Then by the Tate Theorem [12, IV, Theorem 4.7] there exists some maximal subgroup \( V \) of \( P \) such that \( Q\nleq V \), which implies that \( P=QV \) and so \( V\cap R<P\cap R=Q \). If \( V\cap R=1 \), then \( V\cap R=P\cap V\cap R=Q\cap V=1 \) and so \( |Q|=p \); a contradiction. Hence \( V\cap R\neq 1 \). Since \( R=R_{1}\times\cdots\times R_{n} \), where \( R_{1}\simeq\cdots\simeq R_{n} \) are nonabelian simple groups, \( Q=(P\cap R_{1})\times\cdots\times(P\cap R_{n}) \) and so \( V\cap R_{i}<P\cap R_{i} \) for some \( i \). Note also that \( V\cap R_{i}\neq 1 \). Otherwise from the isomorphism

$$ V(P\cap R_{i})/V\simeq(P\cap R_{i})/(V\cap(P\cap R_{i}))=(P\cap R_{i})/1 $$

we get that the order of a Sylow \( p \)-subgroup of \( P\cap R_{i} \) divides \( p \) and so \( P\cap R_{i} \) is \( p \)-nilpotent by [12, IV, Theorem 2.8], which implies that \( R \) is \( p \)-nilpotent.

We show first that \( R \) is \( \sigma \)-primary. Suppose the contrary. We can assume without loss of generality that \( V \) is \( \tau_{\sigma} \)-permutable in \( G \) with respect to \( {\mathcal{H}} \). Then there exists some \( j\neq k \), and for \( H=H_{j} \) we have \( H\cap R_{i}\neq 1 \) because \( R \) is not \( \sigma \)-primary. Note also that \( \sigma_{k}\in\sigma(H^{G}) \). If not, then \( R\cap H^{G}=1 \), which implies that \( 1<H^{G}\leq C_{G}(R)=1 \). Therefore \( \sigma_{k}\in\tau_{\mathcal{H}}(V) \), and so \( VH^{x}=H^{x}V \) for all \( x\in G \). By [13, Chapter A, Lemma 14.1(a)], \( L=VH^{x}\cap R_{i} \) is a subnormal subgroup of \( VH^{x} \), where \( V \) is a Hall \( \sigma_{k} \)-subgroup of \( VH^{x} \) and \( H^{x} \) is a Hall \( \sigma_{j} \)-subgroup of \( VH^{x} \). Therefore, \( L=(L\cap V)(L\cap H^{x}) \) by [13, I, Lemma 3.2]. Hence,

$$ \begin{gathered}\displaystyle L=(L\cap V)(L\cap H^{x})=(VH^{x}\cap R_{i}\cap V)(VH^{x}\cap R_{i}\cap H^{x})\\ \displaystyle=(R_{i}\cap V)(R_{i}\cap H^{x})=(V\cap R_{i})(H\cap R_{i})^{x}=(H\cap R_{i})^{x}(V\cap R_{i})\end{gathered} $$

for all \( x\in R_{i} \), where \( (H\cap R_{i})(V\cap R_{i})\neq R_{i} \) because \( V\cap R_{i}<P\cap R_{i} \). Therefore, \( R_{i} \) is not simple by [8, Lemma 1.1.9(1)] because \( H\cap R_{i}\neq 1 \) and \( V\cap R_{i}\neq 1 \). This contradiction shows that \( R \) is \( \sigma \)-primary.

Then \( H\cap R_{i}\neq 1 \) for some \( j\neq k \) and \( H=H_{j} \). Therefore, \( V\cap R \) is \( \tau_{\sigma} \)-permutable in \( G \) by Lemma 2.6(2). But \( V\cap R\leq R\leq O_{\sigma_{k}}(G) \) and so \( V\cap R \) is \( \sigma \)-permutable in \( G \) by Lemma 2.6(3). Because \( R\leq D\leq O^{\sigma_{i}}(G) \) and so \( R\leq N_{G}(V\cap R) \) by Lemma 2.5, it follows that \( V\cap R\leq O_{p}(R)=1 \); a contradiction. Thus \( R \) is abelian, and so \( D \) is \( p \)-soluble. The lemma is proved.

Lemma 2.10

Let \( D=G^{\mathfrak{N_{\sigma}}} \) and \( \pi=\pi(D) \). If \( {\mathcal{H}}=\{H_{1},\dots,H_{t}\} \) is a complete Hall \( \sigma \)-set of \( G \) such that \( H_{i}\cap D\leq Z_{\mathfrak{U}_{\pi}}(H_{i}) \) for all \( i \), then \( {\mathcal{H}}_{0}=\{H_{1}N/N,\dots,H_{t}N/N\} \) is a complete Hall \( \sigma \)-set of \( G/N \) such that \( (H_{i}N/N)\cap(G/N)^{\mathfrak{N_{\sigma}}}\leq Z_{\mathfrak{U}_{\pi_{0}}}(H_{i}N/N) \) for all \( i \), where \( \pi_{0}=\pi((G/N)^{\mathfrak{N_{\sigma}}}) \).

Proof

It is clear that \( {\mathcal{H}}_{0} \) is a complete Hall \( \sigma \)-set of \( G/N \). Put \( D_{0}=(G/N)^{{\mathfrak{N}}_{\sigma}} \). Then \( D_{0}=DN/N \) by Lemma 2.2, and so

$$ \pi_{0}=\pi(D_{0})=\pi(DN/N)=\pi(D/(D\cap N))\subseteq\pi(D)=\pi. $$

Hence, \( Z_{\mathfrak{U}_{\pi}}(H_{i}N/N)\leq Z_{\mathfrak{U}_{\pi_{0}}}(H_{i}N/N) \). On the other hand, \( D\cap H_{i}N=(D\cap H_{i})(D\cap N) \) by Lemma 2.3. Thus,

$$ D_{0}\cap(H_{i}N/N)=(D\cap H_{i})N/N. $$

Note that, in view of Lemma 2.7(3),

$$ (D\cap H_{i})(N\cap H_{i})/(N\cap H_{i})\leq Z_{\mathfrak{U}_{\pi}}(H_{i}/(N\cap H_{i})) $$

since \( D\cap H_{i}\leq Z_{\mathfrak{U}_{\pi}}(H_{i}) \). Hence

$$ f((D\cap H_{i})(N\cap H_{i})/(N\cap H_{i}))=(D\cap H_{i})N/N\leq Z_{\mathfrak{U}_{\pi}}(H_{i}N/N), $$

where \( f:H_{i}/(N\cap H_{i})\to H_{i}N/N \) is the canonical isomorphism, since

$$ f(Z_{\mathfrak{U}_{\pi}}(H_{i}/(N\cap H_{i})))=Z_{\mathfrak{U}_{\pi}}(H_{i}N/N). $$

Therefore, \( (D\cap H_{i})N/N\leq Z_{\mathfrak{U}_{\pi_{0}}}(H_{i}N/N) \) for all \( i \). The lemma is proved.

Lemma 2.11

Let \( D=G^{\mathfrak{N_{\sigma}}} \) and \( \pi=\pi(D) \). Suppose that \( G \) is \( \sigma \)-soluble and all maximal subgroups of every noncyclic Sylow \( p \)-subgroup of \( G \) are \( \tau_{\sigma} \)-permutable in \( G \) for all \( p\in\pi \). Then

(1) the hypothesis holds for \( G/L \) for every minimal normal subgroup \( L \) of \( G \);

(2) if \( D \) is nilpotent, then \( D \) is a Hall subgroup of \( G \).

Proof

(1): See the proof of Lemma 2.9.

(2): Suppose that this assertion is false. Let \( P \) be a Sylow \( p \)-subgroup of \( D \) and let \( G_{p} \) be a Sylow \( p \)-subgroup of \( G \) such that \( 1<P<G_{p} \). We can assume without loss of generality that \( G_{p}\leq H_{1} \).

(a) \( D=P \) is a minimal normal subgroup of \( G \). Hence \( D\leq G_{p}=H_{1}\trianglelefteq G \).

Let \( R \) be a minimal normal subgroup of \( G \) lying in \( D \). Since \( D \) is nilpotent by hypothesis, \( R \) is a \( q \)-group for some prime \( q \). Moreover, by (1) and the choice of \( G \) we have that \( D/R=(G/R)^{\mathfrak{N}_{\sigma}} \) is a Hall subgroup of \( G/R \). Suppose now that \( PR/R\neq 1 \). Then \( PR/R \) is a Sylow \( p \)-subgroup of \( G/R \). If \( q\neq p \), then \( P \) is a Sylow \( p \)-subgroup of \( G \). This contradicts the fact that \( P<G_{p} \). Hence \( q=p \) and so \( R\leq P \). It implies that \( P/R \) is a Sylow \( p \)-subgroup of \( G/R \), and so \( P \) is a Sylow \( p \)-subgroup of \( G \). This contradiction shows that \( PR/R=1 \), which implies that \( R=P \) is the unique minimal normal subgroup of \( G \) lying in \( D \). Since \( D \) is nilpotent, a \( p^{\prime} \)-complement \( E \) of \( D \) is characteristic in \( D \) and so \( E \) is normal in \( G \). Hence \( E=1 \). This implies that \( R=D=P \). Finally, \( G/D \) is \( \sigma \)-nilpotent by Lemma 2.1 and so \( H_{1}/D \) is normal in \( G/D \). Hence (a) holds.

(b) \( D\nleq\Phi(G) \). Hence there exists a maximal subgroup \( M \) of \( G \) such that \( G=D\rtimes M \). (This follows from (2) and Lemma 2.1 because \( G \) is not \( \sigma \)-nilpotent.)

(c) If \( G \) has a minimal normal subgroup \( L\neq D \), then \( G_{p}=D\times(L\cap G_{p}) \). Hence \( O_{p^{\prime}}(G)=1 \).

By Lemma 2.2, \( (G/L)^{\mathfrak{N}_{\sigma}}=LD/L \). Therefore, by (1), (a), and the choice of \( G \) we have that \( LD/L\simeq D \) is a Hall subgroup of \( G/L \). Hence \( G_{p}L/L=DL/L \), and so \( G_{p}=D\times(L\cap G_{p}) \). Since \( D<G_{p} \) by (a), \( O_{p^{\prime}}(G)=1 \).

(d) \( V=C_{G}(D)\cap M \) is a normal subgroup of \( G \) and \( C_{G}(D)=D\times V\leq H_{1} \).

In view of (a) and (b), \( C_{G}(D)=D\times V \), where \( V=C_{G}(D)\cap M \) is a normal subgroup of \( G \). By (a), \( V\cap D=1 \) and so \( V\simeq DV/D \) is \( \sigma \)-nilpotent by Lemma 2.1. Let \( W \) be a \( \sigma_{1} \)-complement of \( V \). Then \( W \) is characteristic in \( V \) and so it is normal in \( G \). Therefore, (d) holds in view of (c).

(e) \( G_{p}\neq H_{1} \).

Assume that \( G_{p}=H_{1} \). Then \( D<G_{p}\leq C_{G}(D) \) by (a) and [13, Chapter A, Theorem 10.6(b)]. It follows from (d) that \( L\leq C_{G}(D)\cap M\leq G_{p} \) for some minimal normal subgroup \( L \) of \( G \). Hence \( G_{p}=D\times L \) is a normal elementary abelian \( p \)-subgroup of \( G \) by (c). This ensues from Lemmas 2.6(3) and 2.5 that every maximal subgroup of \( G_{p} \) is normal in \( G \). It follows that every subgroup of \( G_{p} \) is normal in \( G \).

Hence \( |D|=|L|=p \). Let \( D=\langle a\rangle \), \( L=\langle b\rangle \), and \( N=\langle ab\rangle \). Then \( N\nleq D \) and so, in view of the \( G \)-isomorphisms

$$ DN/D\simeq N\simeq NL/L=G_{p}/L=DL/L\simeq D, $$

we get that \( G/C_{G}(D)=G/C_{G}(N) \) is a \( p \)-group since \( G_{p}=H_{1} \) and \( G/D \) is \( \sigma \)-nilpotent by Lemma 2.1. It follows from (d) that \( G \) is a \( p \)-group. This contradiction shows that we have (e).

Final contradiction for (2). By Theorem A in [16], \( G \) has a \( \sigma_{1} \)-complement \( E \) such that \( W=EG_{p}=G_{p}E \). Then \( D\leq G_{p}\leq W \) by (a). Moreover, since \( W/D\leq G/D\in{\mathfrak{N_{\sigma}}} \) and \( {\mathfrak{N_{\sigma}}} \) is a hereditary class by Lemma 2.1, \( W/D\in{\mathfrak{N_{\sigma}}} \), and thereby \( V=W^{\mathfrak{N_{\sigma}}}\leq D \). Therefore, in view of Lemmas 2.4 and 2.6(4), the hypothesis holds for \( W \). From (e) we derive that \( W\neq G \). Hence the conclusion of the lemma holds for \( W \) by the choice of \( G \), which implies that \( V \) is a Hall subgroup of \( W \). Moreover, \( V\leq D \) and so \( |V_{p}|\leq|P|<|G_{p}| \) for a Sylow \( p \)-subgroup \( V_{p} \) of \( V \). Hence \( V \) is a \( p^{\prime} \)-group. It implies from (d) that \( V\leq C_{G}(D)\leq H_{1}\cap W \). Therefore \( V=1 \), which shows that \( W=EG_{p}=E\times G_{p} \) is \( \sigma \)-nilpotent and so \( E\leq C_{G}(D)\leq H_{1} \). Hence \( E=1 \). It follows that \( D=1 \), which is a contradiction. Thus \( D \) is a Hall subgroup of \( G \). The lemma is proved.

Proof of Theorem C

Suppose that this theorem is false and let \( G \) be a counterexample of minimal order. Then \( D\neq 1 \). Let \( {\mathcal{H}}=\{H_{1},\dots,H_{t}\} \). We can assume without loss of generality that \( H_{i} \) is a \( \sigma_{i} \)-group for all \( i=1,\dots,t \). Let \( R \) be a minimal normal subgroup of \( G \).

(1) The hypothesis holds for \( G/R \) (see the proof of Lemma 2.9 and use Lemma 2.10).

(2) \( D \) is soluble, and so \( G \) is \( \sigma \)-soluble. Hence \( G \) is a \( \sigma \)-full group of Sylow type (in view of Theorem 2.8 in [12, IV], this follows from Lemmas 2.4, 2.9, and the Feit–Thompson Theorem).

(3) \( D \) is nilpotent.

Assume that this is false. Note that \( (G/R)^{{\mathfrak{N_{\sigma}}}}=RD/R \) is nilpotent by (1) and the choice of \( G \). Therefore \( R\leq D \), while \( R \) is the unique minimal normal subgroup of \( G \) and \( R\nleq\Phi(G) \) by Lemma 2.1. It implies from (2) that \( R \) is a \( p \)-group for some prime \( p \). Therefore, by [13, Chapter A, Theorem 15.2] \( R=C_{G}(R) \), \( G=R\rtimes M \) for some maximal subgroup \( M \) of \( G \) and \( |R|>p \), if not, then \( G/C_{G}(R)=G/R \) is a cyclic group and so \( D \) is nilpotent, contrary to our assumption on \( D \).

Clearly, \( R\leq H_{i}\cap D \) for some \( i \). Then \( H_{i}=R\rtimes(H_{i}\cap M) \) and \( R\leq Z_{\mathfrak{U}_{\pi}}(H_{i}) \) by hypothesis. It shows that there exists a maximal subgroup \( V \) of \( R \) such that \( V \) is normal in \( H_{i} \) because \( p\in\pi \). Let \( P \) be a Sylow \( p \)-subgroup of \( H_{i}\cap M \). Then \( RP \) is a Sylow \( p \)-subgroup of \( G \), and \( VP \) is a maximal subgroup of \( RP \). Hence, by the hypothesis of the theorem \( VP \) is \( \tau_{\sigma} \)-permutable in \( G \). It follows from Lemma 2.6(2)(3) that \( V=V(R\cap P)=R\cap VP \) is \( \sigma \)-permutable in \( G \). Therefore \( O^{\sigma_{i}}(G)\leq N_{G}(V) \) by Lemma 2.5, and thereby \( G=H_{i}O^{\sigma_{i}}(G)\leq N_{G}(V) \). The minimality of \( R \) implies that \( V=1 \) and so \( |R|=p \); a contradiction. Hence, we have (3).

(4) \( D \) is a Hall subgroup of \( G \). (This is straightforward from (2), (3), and Lemma 2.11.)

(5) If \( p \) is a prime such that \( (p-1,|G|)=1 \), then \( p \) does not divide \( |D| \). In particular, the smallest prime divisor of \( |G| \) divides \( |G:D| \).

Assume the contrary and let \( P \) be the Sylow \( p \)-subgroup of \( D \). Then, arguing as in the proof of (3), we can show that some maximal subgroup \( E \) of \( P \) is normal in \( G \). Hence \( C_{G}(D/E)=G \) because \( (p-1,|G|)=1 \) by hypothesis. Since \( D \) is a Hall subgroup of \( G \) by (4), \( D \) has a complement \( M \) in \( G \). Therefore \( G/E=(D/E)\times(ME/E) \), where \( ME/E\simeq M\simeq G/D \) is \( \sigma \)-nilpotent. Thus, \( G/E \) is \( \sigma \)-nilpotent. It follows that \( D\leq E \); a contradiction. Hence \( p \) does not divide \( |D| \). In particular, the smallest prime divisor of \( |G| \) divides \( |G:D| \).

(6) Every chief factor of \( G \) below \( D \) is cyclic.

Suppose the contrary. Assume that \( \Phi(D)\neq 1 \) and let \( R\leq\Phi(D) \). Then the choice of \( G \) and (1) imply that every chief factor of \( G/R \) below \( (G/R)^{{\mathfrak{N}}_{\sigma}}=D/R \) is cyclic, and so every chief factor of \( G \) below \( D \) is cyclic by Lemma 2.8. Hence \( \Phi(D)=1 \), and so every Sylow subgroup of \( D \) is elementary. Moreover, there is \( p\in\pi(D) \) such that the Sylow \( p \)-subgroup \( P \) of \( D \) has a minimal normal subgroup \( N \) of \( G \) such that \( |N|>p \). Let \( V \) be a maximal subgroup of \( P \) such that \( P=NV \). Then \( N\cap V\neq 1 \). Since \( D \) is a Hall subgroup of \( G \), \( P \) is the Sylow \( p \)-subgroup of \( G \). Therefore \( V \) is \( \tau_{\sigma} \)-permutable in \( G \), and so \( N\cap V \) is \( \sigma \)-permutable in \( G \) by Lemma 2.6(2)(3). Arguing as in the proof of (3), we can show that \( N\cap V \) is normal in \( G \). The minimality of \( N \) implies that \( N\cap V=1 \), and so \( |N|=p \). This contradiction completes the proof of (6).

Claims (3)–(6) show that the conclusion of the theorem holds for \( G \), which contradicts the choice of \( G \). The theorem is proved.

3. Proof of Theorem B

Lemma 3.1

Suppose that \( D=G^{{\mathfrak{N}}_{\sigma}} \) is a nilpotent Hall subgroup of \( G \). If every \( \sigma_{i} \)-subgroup of \( G \) is \( \tau_{\sigma} \)-permutable in \( G \) for all \( \sigma_{i}\in\sigma(D) \), then \( D \) is an abelian group of odd order and each element of \( G \) induces a power automorphism in \( D \).

Proof

Suppose that this lemma is false and let \( G \) be a counterexample of minimal order. Let \( {\mathcal{H}}=\{H_{1},\dots,H_{t}\} \). We can assume without loss of generality that \( H_{i} \) is a \( \sigma_{i} \)-group for all \( i=1,\dots,t \).

Note first that

$$ (G/N)^{{\mathfrak{N}}_{\sigma}}=DN/N\simeq D/(D\cap N) $$

is a nilpotent Hall subgroup of \( G/N \) for every minimal normal subgroup \( N \) of \( G \) by Lemma 2.2. Let \( V/N \) be a nonidentity \( \sigma_{i} \)-subgroup of \( G/N \) for some

$$ \sigma_{i}\in\sigma((G/N)^{{\mathfrak{N}}_{\sigma}})=\sigma(DN/N)=\sigma(D/(D\cap N))\subseteq\sigma(D). $$

Let \( U \) be a minimal supplement to \( N \) in \( V \). Then \( U\cap N\leq\Phi(U) \), and so \( U \) is a \( \sigma_{i} \)-subgroup of \( G \) since \( V/N=UN/N\simeq U/(U\cap N) \). Thus, \( U \) is \( \tau_{\sigma} \)-permutable in \( G \) by hypothesis and \( \sigma(U)=\sigma(UN/N)=\{\sigma_{i}\} \), which implies that \( V/N=UN/N \) is \( \tau_{\sigma} \)-permutable in \( G/N \) by Lemma 2.6(1). Hence the hypothesis holds for \( G/N \).

Let \( H \) be a subgroup of the Sylow \( p \)-subgroup \( P \) of \( D \) for some prime \( p\in\pi \). We show that \( H \) is normal in \( G \). For some \( i \) we have \( P\leq O_{\sigma_{i}}(D)=H_{i}\cap D \). On the other hand, \( D=O_{\sigma_{i}}(D)\times O^{\sigma_{i}}(D) \) since \( D \) is nilpotent. Assume that \( O^{\sigma_{i}}(D)\neq 1 \) and let \( N \) be a minimal normal subgroup of \( G \) lying in \( O^{\sigma_{i}}(D) \). Then \( HN/N\leq DN/N=(G/N)^{{\mathfrak{N}}_{\sigma}} \), and so the choice of \( G \) implies that \( HN/N \) is normal in \( G/N \). Hence, \( H=H(N\cap O_{\sigma_{i}}(D))=HN\cap O_{\sigma_{i}}(D) \) is normal in \( G \).

Assume now that \( O^{\sigma_{i}}(D)=1 \). Then \( D \) is a \( \sigma_{i} \)-group. Since \( G/D \) is \( \sigma \)-nilpotent by Lemma 2.1, \( H_{i}/D \) is normal in \( G/D \) and so \( H_{i} \) is normal in \( G \). It follows from Lemma 2.6(3) and the hypothesis of the theorem that all subgroups of \( H_{i} \) are \( \sigma \)-permutable in \( G \). Since \( D \) is a normal Hall subgroup of \( H_{i} \); therefore, \( D \) has a complement \( S \) in \( H_{i} \) by the Schur–Zassenhaus Theorem. It implies from Lemma 2.5 that \( D\leq O^{\sigma_{i}}(G)\leq N_{G}(S) \). Hence \( H_{i}=D\times S \), and so

$$ G=H_{i}O^{\sigma_{i}}(G)=SO^{\sigma_{i}}(G)\leq N_{G}(H). $$

This implies that \( H \) is normal in \( G \). Hence \( D \) is a Dedekind group, and so \( |D| \) is odd by Theorem C. Hence, \( D \) is abelian and each element of \( G \) induces a power automorphism in \( D \). The lemma is proved.

The following lemma is a corollary of Theorem A of this paper and Theorem B in [2].

A subgroup \( A \) of \( G \) is said to be \( {\sigma} \)-subnormal in \( G \) [2] if there is a subgroup chain

$$ A=A_{0}\leq A_{1}\leq\cdots\leq A_{n}=G $$

such that either \( A_{i-1}\trianglelefteq A_{i} \) or \( A_{i}/(A_{i-1})_{A_{i}} \) is \( {\sigma} \)-primary for all \( i=1,\dots,n \).

Lemma 3.2

The following hold:

(i) \( G \) is a \( P\sigma T \)-group if and only if every \( \sigma \)-subnormal subgroup of \( G \) is \( \sigma \)-quasinormal in \( G \).

(ii) If \( G \) is a \( P\sigma T \)-group, then every quotient \( G/N \) of \( G \) is also a \( P\sigma T \)-group.

Proof of Theorem B

Sufficiency: Assume the contrary and let \( G \) be a counterexample with \( |G|+|N| \) minimal. By Lemma 2.1, \( D:=G^{\mathfrak{N}_{\sigma}} \) is the smallest normal subgroup of \( G \) with \( \sigma \)-nilpotent quotient. Therefore \( D\leq N \) and so the hypothesis holds for \( (G,D) \). Hence the choice of \( G \) shows that \( D=N \). We can assume without loss of generality that \( H_{i} \) is a \( \sigma_{i} \)-group for all \( i=1,\dots,t \).

(1) \( G=D\rtimes M \), where \( D \) is an abelian Hall subgroup of \( G \) of odd order, \( M \) is \( \sigma \)-nilpotent, and every element of \( G \) induces a power automorphism in \( D \). (This is straightforward from Lemma 3.1 and Theorem C.)

(2) If \( R \) is a nonidentity normal subgroup of \( G \), then the hypothesis holds for \( G/R \), and so \( G/R \) is a \( \sigma \)-soluble \( P\sigma T \)-group (see the proof of Lemma 3.1 and use Lemma 2.10).

(3) \( H_{i}=O_{\sigma_{i}}(D)\times S \) for some subgroup \( S \) of \( H_{i} \) for each \( \sigma_{i}\in\sigma(D) \).

Since \( D \) is an abelian Hall subgroup of \( G \) by (1), \( D=L\times N \), where \( L=O_{\sigma_{i}}(D) \) and \( N=O^{\sigma_{i}}(D)=O_{\sigma^{\prime}_{i}}(D) \) are Hall subgroups of \( G \). Assume first that \( N\neq 1 \). Then

$$ O_{\sigma_{i}}((G/N)^{\mathfrak{N}_{\sigma}})=O_{\sigma_{i}}(D/N)=LN/N $$

has a normal complement \( V/N \) in \( H_{i}N/N\simeq H_{i} \) by (2) and Theorem A. On the other hand, \( N \) has a complement \( S \) in \( V \) by the Schur–Zassenhaus Theorem. Hence \( H_{i}=H_{i}\cap LSN=LS \) and \( L\cap S=1 \) since

$$ (L\cap S)N/N\leq(LN/N)\cap(V/N)=(LN/N)\cap(SN/N)=1. $$

It is clear that \( V/N \) is a Hall subgroup of \( H_{i}N/N \), and so \( V/N \) is characteristic in \( H_{i}N/N \). On the other hand, \( H_{i}N/N \) is normal in \( G/N \) by Lemma 2.2 since \( D/N\leq H_{i}N/N \). Hence \( V/N \) is normal in \( G/N \). Thus \( H_{i}\cap V=H_{i}\cap NS=S(H_{i}\cap N)=S \) is normal in \( H_{i} \), and so \( H_{i}=O_{\sigma_{i}}(D)\times S \).

Assume that \( D=O_{\sigma_{i}}(D) \). Then \( H_{i} \) is normal in \( G \), and so all subgroups of \( H_{i} \) are \( \sigma \)-permutable in \( G \) by Lemma 2.6(3). Since \( D \) is a normal Hall subgroup of \( H_{i} \), \( D \) has a complement \( S \) in \( H_{i} \). Using Lemma 2.5, we imply that \( D\leq O^{\sigma_{i}}(G)\leq N_{G}(S) \). Hence, \( H_{i}=D\times S=O_{\sigma_{i}}(D)\times S \).

It follows from Theorem A, (2), and (3) that \( G \) is a \( \sigma \)-soluble \( P\sigma T \)-group, contrary to our assumption on \( G \). This completes the proof of sufficiency.

Assume now that \( G \) is a \( \sigma \)-soluble \( P\sigma T \)-group and let \( D=G^{\mathfrak{N}_{\sigma}} \). Then \( G \) possesses a complete \( \sigma \)-set \( {\mathcal{H}}=\{H_{1},\dots,H_{t}\} \) by Lemma 2.4. Moreover, \( G/D \) is \( \sigma \)-nilpotent by Lemma 2.1 and every subgroup of \( D \) is normal in \( G \) by Theorem A. Then \( H_{i}\cap N\leq Z_{\mathfrak{U}}(H_{i})\leq Z_{\mathfrak{U}_{\pi}}(H_{i}) \), where \( \pi=\pi(N) \) for all \( i \). Therefore, (i) holds for \( G \).

We show now that every \( \sigma_{i} \)-subgroup of \( G \) is \( \tau_{\sigma} \)-permutable in \( G \) for each \( \sigma_{i}\in\sigma(D) \). It suffices to show that if \( H \) is a \( \sigma_{i} \)-subgroup of \( G \), and so \( H \) permutes with every Hall \( \sigma_{j} \)-subgroup of \( G \) for all \( j\neq i \). Assume the contrary and let \( G \) be a counterexample of minimal order. Then \( D\neq 1 \) and there are \( \sigma_{i} \) and \( \sigma_{j} \) (\( i\neq j \)) such that \( \sigma_{i}\in\sigma(D) \) and \( HE\neq EH \) for some \( \sigma_{i} \)-subgroup \( H \) and some Hall \( \sigma_{j} \)-subgroup \( E \) of \( G \). Then \( H \) is not \( \sigma \)-subnormal in \( G \) by Lemma 3.2. Hence a Hall \( \sigma_{i} \)-subgroup \( H_{i} \) of \( G \) is not normal in \( G \) since otherwise \( H\leq H_{i} \) and so \( H \) is \( \sigma \)-subnormal in \( G \) by Lemma 2.6. Note that \( |\sigma(D)|>1 \). Indeed, if \( |\sigma(D)|=1 \), then \( \sigma(D)=\{\sigma_{i}\} \) and so \( D\leq H_{i} \), which implies that \( H_{i}/D \) is normal in \( G/D \) because \( G/D \) is \( \sigma \)-nilpotent. Hence \( H_{i} \) is normal in \( G \); a contradiction.

We show now that \( EHN \) is a subgroup of \( G \) for every minimal normal subgroup \( N \) of \( G \). Note first that the hypothesis holds for \( G/N \) by Lemma 3.2. Moreover, \( HN/N\simeq H/(H\cap N) \) is a \( \sigma_{i} \)-subgroup of \( G/N \). Therefore, if \( \sigma_{i}\in\sigma(DN/N)=\sigma((G/N)^{{\mathfrak{N}}_{\sigma}}) \), then the choice of \( G \) implies that

$$ (HN/N)(EN/N)=(EN/N)(HN/N)=EHN/N $$

is a subgroup of \( G/N \). Hence \( EHN \) is a subgroup of \( G \). Assume now that \( \sigma_{i}\not\in\sigma(DN/N) \). Then a Hall \( \sigma_{i} \)-subgroup \( H_{i} \) of \( G \) lies in \( N \). Clearly, \( H_{i}=N \) because \( N \) is \( \sigma \)-primary. It follows that \( H\leq N \) and so \( H \) is \( \sigma \)-subnormal in \( G \); a contradiction. Hence \( EHN \) is a subgroup of \( G \). Since \( |\sigma(D)|>1 \) and \( D \) is abelian by Theorem A, \( G \) has at least two \( \sigma \)-primary minimal normal subgroups \( R \) and \( N \) such that \( R,N\leq D \) and \( \sigma(R)\neq\sigma(N) \). Then at least one of the subgroups \( R \) or \( N \), say \( R \), is a \( \sigma_{k} \)-group for some \( k\neq j \). Moreover,

$$ R\cap E(HN)=(R\cap E)(R\cap HN)=R\cap HN $$

by Lemma 2.3 and \( R\cap HN\leq O_{\sigma_{k}}(HN)\leq V \), where \( V \) is a Hall \( \sigma_{k} \)-subgroup of \( H \), because \( N \) is a \( \sigma_{k}^{\prime} \)-group and \( G \) is a \( \sigma \)-full group of Sylow type by Lemma 2.4. Hence

$$ EHR\cap EHN=EH(R\cap E(HN))=EH(R\cap HN)=EH(R\cap H)=EH $$

is a subgroup of \( G \), i.e., \( HE=EH \). This contradicts \( HE\neq EH \). Therefore, (ii) holds for \( G \). Hence the necessity of the condition of the theorem holds for \( G \). The theorem is proved.